Datasets:

CShorten

/

Last-Week-on-ML-ArXiv

like 0

SLOVA: Uncertainty Estimation Using Single Label One-Vs-All Classifier

Deep neural networks present impressive performance, yet they cannot reliably estimate their predictive confidence, limiting their applicability in high-risk domains. We show that applying a multi-label one-vs-all loss reveals classification ambiguity and reduces model overconfidence. The introduced SLOVA (Single Label One-Vs-All) model redefines typical one-vs-all predictive probabilities to a single label situation, where only one class is the correct answer. The proposed classifier is confident only if a single class has a high probability and other probabilities are negligible. Unlike the typical softmax function, SLOVA naturally detects out-of-distribution samples if the probabilities of all other classes are small. The model is additionally fine-tuned with exponential calibration, which allows us to precisely align the confidence score with model accuracy. We verify our approach on three tasks. First, we demonstrate that SLOVA is competitive with the state-of-the-art on in-distribution calibration. Second, the performance of SLOVA is robust under dataset shifts. Finally, our approach performs extremely well in the detection of out-of-distribution samples. Consequently, SLOVA is a tool that can be used in various applications where uncertainty modeling is required.

QTI Submission to DCASE 2021: residual normalization for device-imbalanced acoustic scene classification with efficient design

This technical report describes the details of our TASK1A submission of the DCASE2021 challenge. The goal of the task is to design an audio scene classification system for device-imbalanced datasets under the constraints of model complexity. This report introduces four methods to achieve the goal. First, we propose Residual Normalization, a novel feature normalization method that uses instance normalization with a shortcut path to discard unnecessary device-specific information without losing useful information for classification. Second, we design an efficient architecture, BC-ResNet-Mod, a modified version of the baseline architecture with a limited receptive field. Third, we exploit spectrogram-to-spectrogram translation from one to multiple devices to augment training data. Finally, we utilize three model compression schemes: pruning, quantization, and knowledge distillation to reduce model complexity. The proposed system achieves an average test accuracy of 76.3% in TAU Urban Acoustic Scenes 2020 Mobile, development dataset with 315k parameters, and average test accuracy of 75.3% after compression to 61.0KB of non-zero parameters.

Feature Learning for Dimensionality Reduction toward Maximal Extraction of Hidden Patterns

Dimensionality reduction (DR) plays a vital role in the visual analysis of high-dimensional data. One main aim of DR is to reveal hidden patterns that lie on intrinsic low-dimensional manifolds. However, DR often overlooks important patterns when the manifolds are strongly distorted or hidden by certain influential data attributes. This paper presents a feature learning framework, FEALM, designed to generate an optimized set of data projections for nonlinear DR in order to capture important patterns in the hidden manifolds. These projections produce maximally different nearest-neighbor graphs so that resultant DR outcomes are significantly different. To achieve such a capability, we design an optimization algorithm as well as introduce a new graph dissimilarity measure, called neighbor-shape dissimilarity. Additionally, we develop interactive visualizations to assist comparison of obtained DR results and interpretation of each DR result. We demonstrate FEALM's effectiveness through experiments using synthetic datasets and multiple case studies on real-world datasets.

Parallel Instance Filtering for Malware Detection

Machine learning algorithms are widely used in the area of malware detection. With the growth of sample amounts, training of classification algorithms becomes more and more expensive. In addition, training data sets may contain redundant or noisy instances. The problem to be solved is how to select representative instances from large training data sets without reducing the accuracy. This work presents a new parallel instance selection algorithm called Parallel Instance Filtering (PIF). The main idea of the algorithm is to split the data set into non-overlapping subsets of instances covering the whole data set and apply a filtering process for each subset. Each subset consists of instances that have the same nearest enemy. As a result, the PIF algorithm is fast since subsets are processed independently of each other using parallel computation. We compare the PIF algorithm with several state-of-the-art instance selection algorithms on a large data set of 500,000 malicious and benign samples. The feature set was extracted using static analysis, and it includes metadata from the portable executable file format. Our experimental results demonstrate that the proposed instance selection algorithm reduces the size of a training data set significantly with the only slightly decreased accuracy. The PIF algorithm outperforms existing instance selection methods used in the experiments in terms of the ratio between average classification accuracy and storage percentage.

Explaining Any ML Model? -- On Goals and Capabilities of XAI

An increasing ubiquity of machine learning (ML) motivates research on algorithms to explain ML models and their predictions -- so-called eXplainable Artificial Intelligence (XAI). Despite many survey papers and discussions, the goals and capabilities of XAI algorithms are far from being well understood. We argue that this is because of a problematic reasoning scheme in XAI literature: XAI algorithms are said to complement ML models with desired properties, such as "interpretability", or "explainability". These properties are in turn assumed to contribute to a goal, like "trust" in an ML system. But most properties lack precise definitions and their relationship to such goals is far from obvious. The result is a reasoning scheme that obfuscates research results and leaves an important question unanswered: What can one expect from XAI algorithms? In this article, we clarify the goals and capabilities of XAI algorithms from a concrete perspective: that of their users. Explaining ML models is only necessary if users have questions about them. We show that users can ask diverse questions, but that only one of them can be answered by current XAI algorithms. Answering this core question can be trivial, difficult or even impossible, depending on the ML application. Based on these insights, we outline which capabilities policymakers, researchers and society can reasonably expect from XAI algorithms.

Disentangling Embedding Spaces with Minimal Distributional Assumptions

Interest in understanding and factorizing learned embedding spaces is growing. For instance, recent concept-based explanation techniques analyze a machine learning model in terms of interpretable latent components. Such components have to be discovered in the model's embedding space, e.g., through independent component analysis (ICA) or modern disentanglement learning techniques. While these unsupervised approaches offer a sound formal framework, they either require access to a data generating function or impose rigid assumptions on the data distribution, such as independence of components, that are often violated in practice. In this work, we link conceptual explainability for vision models with disentanglement learning and ICA. This enables us to provide first theoretical results on how components can be identified without requiring any distributional assumptions. From these insights, we derive the disjoint attributions (DA) concept discovery method that is applicable to a broader class of problems than current approaches but yet possesses a formal identifiability guarantee. In an extensive comparison against component analysis and over 300 state-of-the-art disentanglement models, DA stably maintains superior performance, even under varying distributions and correlation strengths.

LiteCON: An All-Photonic Neuromorphic Accelerator for Energy-efficient Deep Learning (Preprint)

Deep learning is highly pervasive in today's data-intensive era. In particular, convolutional neural networks (CNNs) are being widely adopted in a variety of fields for superior accuracy. However, computing deep CNNs on traditional CPUs and GPUs brings several performance and energy pitfalls. Several novel approaches based on ASIC, FPGA, and resistive-memory devices have been recently demonstrated with promising results. Most of them target only the inference (testing) phase of deep learning. There have been very limited attempts to design a full-fledged deep learning accelerator capable of both training and inference. It is due to the highly compute and memory-intensive nature of the training phase. In this paper, we propose LiteCON, a novel analog photonics CNN accelerator. LiteCON uses silicon microdisk-based convolution, memristor-based memory, and dense-wavelength-division-multiplexing for energy-efficient and ultrafast deep learning. We evaluate LiteCON using a commercial CAD framework (IPKISS) on deep learning benchmark models including LeNet and VGG-Net. Compared to the state-of-the-art, LiteCON improves the CNN throughput, energy efficiency, and computational efficiency by up to 32x, 37x, and 5x respectively with trivial accuracy degradation.

Learning the Evolutionary and Multi-scale Graph Structure for Multivariate Time Series Forecasting

Recent studies have shown great promise in applying graph neural networks for multivariate time series forecasting, where the interactions of time series are described as a graph structure and the variables are represented as the graph nodes. Along this line, existing methods usually assume that the graph structure (or the adjacency matrix), which determines the aggregation manner of graph neural network, is fixed either by definition or self-learning. However, the interactions of variables can be dynamic and evolutionary in real-world scenarios. Furthermore, the interactions of time series are quite different if they are observed at different time scales. To equip the graph neural network with a flexible and practical graph structure, in this paper, we investigate how to model the evolutionary and multi-scale interactions of time series. In particular, we first provide a hierarchical graph structure cooperated with the dilated convolution to capture the scale-specific correlations among time series. Then, a series of adjacency matrices are constructed under a recurrent manner to represent the evolving correlations at each layer. Moreover, a unified neural network is provided to integrate the components above to get the final prediction. In this way, we can capture the pair-wise correlations and temporal dependency simultaneously. Finally, experiments on both single-step and multi-step forecasting tasks demonstrate the superiority of our method over the state-of-the-art approaches.

Sublinear-Time Clustering Oracle for Signed Graphs

Social networks are often modeled using signed graphs, where vertices correspond to users and edges have a sign that indicates whether an interaction between users was positive or negative. The arising signed graphs typically contain a clear community structure in the sense that the graph can be partitioned into a small number of polarized communities, each defining a sparse cut and indivisible into smaller polarized sub-communities. We provide a local clustering oracle for signed graphs with such a clear community structure, that can answer membership queries, i.e., "Given a vertex $v$, which community does $v$ belong to?", in sublinear time by reading only a small portion of the graph. Formally, when the graph has bounded maximum degree and the number of communities is at most $O(\log n)$, then with $\tilde{O}(\sqrt{n}\operatorname{poly}(1/\varepsilon))$ preprocessing time, our oracle can answer each membership query in $\tilde{O}(\sqrt{n}\operatorname{poly}(1/\varepsilon))$ time, and it correctly classifies a $(1-\varepsilon)$-fraction of vertices w.r.t. a set of hidden planted ground-truth communities. Our oracle is desirable in applications where the clustering information is needed for only a small number of vertices. Previously, such local clustering oracles were only known for unsigned graphs; our generalization to signed graphs requires a number of new ideas and gives a novel spectral analysis of the behavior of random walks with signs. We evaluate our algorithm for constructing such an oracle and answering membership queries on both synthetic and real-world datasets, validating its performance in practice.

Improving Correlation Capture in Generating Imbalanced Data using Differentially Private Conditional GANs

Despite the remarkable success of Generative Adversarial Networks (GANs) on text, images, and videos, generating high-quality tabular data is still under development owing to some unique challenges such as capturing dependencies in imbalanced data, optimizing the quality of synthetic patient data while preserving privacy. In this paper, we propose DP-CGANS, a differentially private conditional GAN framework consisting of data transformation, sampling, conditioning, and networks training to generate realistic and privacy-preserving tabular data. DP-CGANS distinguishes categorical and continuous variables and transforms them to latent space separately. Then, we structure a conditional vector as an additional input to not only presents the minority class in the imbalanced data, but also capture the dependency between variables. We inject statistical noise to the gradients in the networking training process of DP-CGANS to provide a differential privacy guarantee. We extensively evaluate our model with state-of-the-art generative models on three public datasets and two real-world personal health datasets in terms of statistical similarity, machine learning performance, and privacy measurement. We demonstrate that our model outperforms other comparable models, especially in capturing dependency between variables. Finally, we present the balance between data utility and privacy in synthetic data generation considering the different data structure and characteristics of real-world datasets such as imbalance variables, abnormal distributions, and sparsity of data.

Detecting Arbitrary Order Beneficial Feature Interactions for Recommender Systems

Detecting beneficial feature interactions is essential in recommender systems, and existing approaches achieve this by examining all the possible feature interactions. However, the cost of examining all the possible higher-order feature interactions is prohibitive (exponentially growing with the order increasing). Hence existing approaches only detect limited order (e.g., combinations of up to four features) beneficial feature interactions, which may miss beneficial feature interactions with orders higher than the limitation. In this paper, we propose a hypergraph neural network based model named HIRS. HIRS is the first work that directly generates beneficial feature interactions of arbitrary orders and makes recommendation predictions accordingly. The number of generated feature interactions can be specified to be much smaller than the number of all the possible interactions and hence, our model admits a much lower running time. To achieve an effective algorithm, we exploit three properties of beneficial feature interactions, and propose deep-infomax-based methods to guide the interaction generation. Our experimental results show that HIRS outperforms state-of-the-art algorithms by up to 5% in terms of recommendation accuracy.

Classification of ADHD Patients Using Kernel Hierarchical Extreme Learning Machine

Recently, the application of deep learning models to diagnose neuropsychiatric diseases from brain imaging data has received more and more attention. However, in practice, exploring interactions in brain functional connectivity based on operational magnetic resonance imaging data is critical for studying mental illness. Since Attention-Deficit and Hyperactivity Disorder (ADHD) is a type of chronic disease that is very difficult to diagnose in the early stages, it is necessary to improve the diagnosis accuracy of such illness using machine learning models treating patients before the critical condition. In this study, we utilize the dynamics of brain functional connectivity to model features from medical imaging data, which can extract the differences in brain function interactions between Normal Control (NC) and ADHD. To meet that requirement, we employ the Bayesian connectivity change-point model to detect brain dynamics using the local binary encoding approach and kernel hierarchical extreme learning machine for classifying features. To verify our model, we experimented with it on several real-world children's datasets, and our results achieved superior classification rates compared to the state-of-the-art models.

Exploring linguistic feature and model combination for speech recognition based automatic AD detection

Early diagnosis of Alzheimer's disease (AD) is crucial in facilitating preventive care and delay progression. Speech based automatic AD screening systems provide a non-intrusive and more scalable alternative to other clinical screening techniques. Scarcity of such specialist data leads to uncertainty in both model selection and feature learning when developing such systems. To this end, this paper investigates the use of feature and model combination approaches to improve the robustness of domain fine-tuning of BERT and Roberta pre-trained text encoders on limited data, before the resulting embedding features being fed into an ensemble of backend classifiers to produce the final AD detection decision via majority voting. Experiments conducted on the ADReSS20 Challenge dataset suggest consistent performance improvements were obtained using model and feature combination in system development. State-of-the-art AD detection accuracies of 91.67 percent and 93.75 percent were obtained using manual and ASR speech transcripts respectively on the ADReSS20 test set consisting of 48 elderly speakers.

ECG Heartbeat classification using deep transfer learning with Convolutional Neural Network and STFT technique

Electrocardiogram (ECG) is a simple non-invasive measure to identify heart-related issues such as irregular heartbeats known as arrhythmias. While artificial intelligence and machine learning is being utilized in a wide range of healthcare related applications and datasets, many arrhythmia classifiers using deep learning methods have been proposed in recent years. However, sizes of the available datasets from which to build and assess machine learning models is often very small and the lack of well-annotated public ECG datasets is evident. In this paper, we propose a deep transfer learning framework that is aimed to perform classification on a small size training dataset. The proposed method is to fine-tune a general-purpose image classifier ResNet-18 with MIT-BIH arrhythmia dataset in accordance with the AAMI EC57 standard. This paper further investigates many existing deep learning models that have failed to avoid data leakage against AAMI recommendations. We compare how different data split methods impact the model performance. This comparison study implies that future work in arrhythmia classification should follow the AAMI EC57 standard when using any including MIT-BIH arrhythmia dataset.

DistSPECTRL: Distributing Specifications in Multi-Agent Reinforcement Learning Systems

While notable progress has been made in specifying and learning objectives for general cyber-physical systems, applying these methods to distributed multi-agent systems still pose significant challenges. Among these are the need to (a) craft specification primitives that allow expression and interplay of both local and global objectives, (b) tame explosion in the state and action spaces to enable effective learning, and (c) minimize coordination frequency and the set of engaged participants for global objectives. To address these challenges, we propose a novel specification framework that allows natural composition of local and global objectives used to guide training of a multi-agent system. Our technique enables learning expressive policies that allow agents to operate in a coordination-free manner for local objectives, while using a decentralized communication protocol for enforcing global ones. Experimental results support our claim that sophisticated multi-agent distributed planning problems can be effectively realized using specification-guided learning.

Adaptive Multi-view Rule Discovery for Weakly-Supervised Compatible Products Prediction

On e-commerce platforms, predicting if two products are compatible with each other is an important functionality to achieve trustworthy product recommendation and search experience for consumers. However, accurately predicting product compatibility is difficult due to the heterogeneous product data and the lack of manually curated training data. We study the problem of discovering effective labeling rules that can enable weakly-supervised product compatibility prediction. We develop AMRule, a multi-view rule discovery framework that can (1) adaptively and iteratively discover novel rulers that can complement the current weakly-supervised model to improve compatibility prediction; (2) discover interpretable rules from both structured attribute tables and unstructured product descriptions. AMRule adaptively discovers labeling rules from large-error instances via a boosting-style strategy, the high-quality rules can remedy the current model's weak spots and refine the model iteratively. For rule discovery from structured product attributes, we generate composable high-order rules from decision trees; and for rule discovery from unstructured product descriptions, we generate prompt-based rules from a pre-trained language model. Experiments on 4 real-world datasets show that AMRule outperforms the baselines by 5.98% on average and improves rule quality and rule proposal efficiency.

Cooperative Retriever and Ranker in Deep Recommenders

Deep recommender systems jointly leverage the retrieval and ranking operations to generate the recommendation result. The retriever targets selecting a small set of relevant candidates from the entire items with high efficiency; while the ranker, usually more precise but time-consuming, is supposed to identify the best items out of the retrieved candidates with high precision. However, the retriever and ranker are usually trained in poorly-cooperative ways, leading to limited recommendation performances when working as an entirety. In this work, we propose a novel DRS training framework CoRR(short for Cooperative Retriever and Ranker), where the retriever and ranker can be mutually reinforced. On one hand, the retriever is learned from recommendation data and the ranker via knowledge distillation; knowing that the ranker is more precise, the knowledge distillation may provide extra weak-supervision signals for the improvement of retrieval quality. On the other hand, the ranker is trained by learning to discriminate the truth positive items from hard negative candidates sampled from the retriever. With the iteration going on, the ranker may become more precise, which in return gives rise to informative training signals for the retriever; meanwhile, with the improvement of retriever, harder negative candidates can be sampled, which contributes to a higher discriminative capability of the ranker. To facilitate the effective conduct of CoRR, an asymptotic-unbiased approximation of KL divergence is introduced for the knowledge distillation over sampled items; besides, a scalable and adaptive strategy is developed to efficiently sample from the retriever. Comprehensive experimental studies are performed over four large-scale benchmark datasets, where CoRR improves the overall recommendation quality resulting from the cooperation between retriever and ranker.

H-GCN: A Graph Convolutional Network Accelerator on Versal ACAP Architecture

Graph Neural Networks (GNNs) have drawn tremendous attention due to their unique capability to extend Machine Learning (ML) approaches to applications broadly-defined as having unstructured data, especially graphs. Compared with other Machine Learning (ML) modalities, the acceleration of Graph Neural Networks (GNNs) is more challenging due to the irregularity and heterogeneity derived from graph typologies. Existing efforts, however, have focused mainly on handling graphs' irregularity and have not studied their heterogeneity. To this end we propose H-GCN, a PL (Programmable Logic) and AIE (AI Engine) based hybrid accelerator that leverages the emerging heterogeneity of Xilinx Versal Adaptive Compute Acceleration Platforms (ACAPs) to achieve high-performance GNN inference. In particular, H-GCN partitions each graph into three subgraphs based on its inherent heterogeneity, and processes them using PL and AIE, respectively. To further improve performance, we explore the sparsity support of AIE and develop an efficient density-aware method to automatically map tiles of sparse matrix-matrix multiplication (SpMM) onto the systolic tensor array. Compared with state-of-the-art GCN accelerators, H-GCN achieves, on average, speedups of 1.1~2.3X.

Persistent homology-based descriptor for machine-learning potential

Constructing efficient descriptors that represent atomic configurations is crucial for developing a superior machine-learning potential. Widely used conventional descriptors are based on two- or three-body correlations of atomic distribution. Recently, several limitations of these many-body descriptors in classifying different configurations were revealed, which have detrimental effects on the prediction of physical properties. We proposed a new class of descriptors based on persistent homology. We focused on the two-dimensional visualization of persistent homology, that is, a persistence diagram, as a descriptor of atomic configurations in the form of an image. We demonstrated that convolutional neural network models based on this descriptor provide sufficient accuracy in predicting the mean energies per atom of amorphous graphene and amorphous carbon. Our results provide an avenue for improving machine-learning potential using descriptors that depict both topological and geometric information.

Secure Forward Aggregation for Vertical Federated Neural Networks

Vertical federated learning (VFL) is attracting much attention because it enables cross-silo data cooperation in a privacy-preserving manner. While most research works in VFL focus on linear and tree models, deep models (e.g., neural networks) are not well studied in VFL. In this paper, we focus on SplitNN, a well-known neural network framework in VFL, and identify a trade-off between data security and model performance in SplitNN. Briefly, SplitNN trains the model by exchanging gradients and transformed data. On the one hand, SplitNN suffers from the loss of model performance since multiply parties jointly train the model using transformed data instead of raw data, and a large amount of low-level feature information is discarded. On the other hand, a naive solution of increasing the model performance through aggregating at lower layers in SplitNN (i.e., the data is less transformed and more low-level feature is preserved) makes raw data vulnerable to inference attacks. To mitigate the above trade-off, we propose a new neural network protocol in VFL called Security Forward Aggregation (SFA). It changes the way of aggregating the transformed data and adopts removable masks to protect the raw data. Experiment results show that networks with SFA achieve both data security and high model performance.

SHELS: Exclusive Feature Sets for Novelty Detection and Continual Learning Without Class Boundaries

While deep neural networks (DNNs) have achieved impressive classification performance in closed-world learning scenarios, they typically fail to generalize to unseen categories in dynamic open-world environments, in which the number of concepts is unbounded. In contrast, human and animal learners have the ability to incrementally update their knowledge by recognizing and adapting to novel observations. In particular, humans characterize concepts via exclusive (unique) sets of essential features, which are used for both recognizing known classes and identifying novelty. Inspired by natural learners, we introduce a Sparse High-level-Exclusive, Low-level-Shared feature representation (SHELS) that simultaneously encourages learning exclusive sets of high-level features and essential, shared low-level features. The exclusivity of the high-level features enables the DNN to automatically detect out-of-distribution (OOD) data, while the efficient use of capacity via sparse low-level features permits accommodating new knowledge. The resulting approach uses OOD detection to perform class-incremental continual learning without known class boundaries. We show that using SHELS for novelty detection results in statistically significant improvements over state-of-the-art OOD detection approaches over a variety of benchmark datasets. Further, we demonstrate that the SHELS model mitigates catastrophic forgetting in a class-incremental learning setting,enabling a combined novelty detection and accommodation framework that supports learning in open-world settings

Generalized Policy Improvement Algorithms with Theoretically Supported Sample Reuse

Real-world sequential decision making requires data-driven algorithms that provide practical guarantees on performance throughout training while also making efficient use of data. Model-free deep reinforcement learning represents a framework for such data-driven decision making, but existing algorithms typically only focus on one of these goals while sacrificing performance with respect to the other. On-policy algorithms guarantee policy improvement throughout training but suffer from high sample complexity, while off-policy algorithms make efficient use of data through sample reuse but lack theoretical guarantees. In order to balance these competing goals, we develop a class of Generalized Policy Improvement algorithms that combines the policy improvement guarantees of on-policy methods with the efficiency of theoretically supported sample reuse. We demonstrate the benefits of this new class of algorithms through extensive experimental analysis on a variety of continuous control tasks from the DeepMind Control Suite.

Personalized Keyword Spotting through Multi-task Learning

Keyword spotting (KWS) plays an essential role in enabling speech-based user interaction on smart devices, and conventional KWS (C-KWS) approaches have concentrated on detecting user-agnostic pre-defined keywords. However, in practice, most user interactions come from target users enrolled in the device which motivates to construct personalized keyword spotting. We design two personalized KWS tasks; (1) Target user Biased KWS (TB-KWS) and (2) Target user Only KWS (TO-KWS). To solve the tasks, we propose personalized keyword spotting through multi-task learning (PK-MTL) that consists of multi-task learning and task-adaptation. First, we introduce applying multi-task learning on keyword spotting and speaker verification to leverage user information to the keyword spotting system. Next, we design task-specific scoring functions to adapt to the personalized KWS tasks thoroughly. We evaluate our framework on conventional and personalized scenarios, and the results show that PK-MTL can dramatically reduce the false alarm rate, especially in various practical scenarios.

Traffic Management of Autonomous Vehicles using Policy Based Deep Reinforcement Learning and Intelligent Routing

Deep Reinforcement Learning (DRL) uses diverse, unstructured data and makes RL capable of learning complex policies in high dimensional environments. Intelligent Transportation System (ITS) based on Autonomous Vehicles (AVs) offers an excellent playground for policy-based DRL. Deep learning architectures solve computational challenges of traditional algorithms while helping in real-world adoption and deployment of AVs. One of the main challenges in AVs implementation is that it can worsen traffic congestion on roads if not reliably and efficiently managed. Considering each vehicle's holistic effect and using efficient and reliable techniques could genuinely help optimise traffic flow management and congestion reduction. For this purpose, we proposed a intelligent traffic control system that deals with complex traffic congestion scenarios at intersections and behind the intersections. We proposed a DRL-based signal control system that dynamically adjusts traffic signals according to the current congestion situation on intersections. To deal with the congestion on roads behind the intersection, we used re-routing technique to load balance the vehicles on road networks. To achieve the actual benefits of the proposed approach, we break down the data silos and use all the data coming from sensors, detectors, vehicles and roads in combination to achieve sustainable results. We used SUMO micro-simulator for our simulations. The significance of our proposed approach is manifested from the results.

Domain Agnostic Few-shot Learning for Speaker Verification

Deep learning models for verification systems often fail to generalize to new users and new environments, even though they learn highly discriminative features. To address this problem, we propose a few-shot domain generalization framework that learns to tackle distribution shift for new users and new domains. Our framework consists of domain-specific and domain-aggregation networks, which are the experts on specific and combined domains, respectively. By using these networks, we generate episodes that mimic the presence of both novel users and novel domains in the training phase to eventually produce better generalization. To save memory, we reduce the number of domain-specific networks by clustering similar domains together. Upon extensive evaluation on artificially generated noise domains, we can explicitly show generalization ability of our framework. In addition, we apply our proposed methods to the existing competitive architecture on the standard benchmark, which shows further performance improvements.

Graph Condensation via Receptive Field Distribution Matching

Graph neural networks (GNNs) enable the analysis of graphs using deep learning, with promising results in capturing structured information in graphs. This paper focuses on creating a small graph to represent the original graph, so that GNNs trained on the size-reduced graph can make accurate predictions. We view the original graph as a distribution of receptive fields and aim to synthesize a small graph whose receptive fields share a similar distribution. Thus, we propose Graph Condesation via Receptive Field Distribution Matching (GCDM), which is accomplished by optimizing the synthetic graph through the use of a distribution matching loss quantified by maximum mean discrepancy (MMD). Additionally, we demonstrate that the synthetic graph generated by GCDM is highly generalizable to a variety of models in evaluation phase and that the condensing speed is significantly improved using this framework.

Dummy Prototypical Networks for Few-Shot Open-Set Keyword Spotting

Keyword spotting is the task of detecting a keyword in streaming audio. Conventional keyword spotting targets predefined keywords classification, but there is growing attention in few-shot (query-by-example) keyword spotting, e.g., N-way classification given M-shot support samples. Moreover, in real-world scenarios, there can be utterances from unexpected categories (open-set) which need to be rejected rather than classified as one of the N classes. Combining the two needs, we tackle few-shot open-set keyword spotting with a new benchmark setting, named splitGSC. We propose episode-known dummy prototypes based on metric learning to detect an open-set better and introduce a simple and powerful approach, Dummy Prototypical Networks (D-ProtoNets). Our D-ProtoNets shows clear margins compared to recent few-shot open-set recognition (FSOSR) approaches in the suggested splitGSC. We also verify our method on a standard benchmark, miniImageNet, and D-ProtoNets shows the state-of-the-art open-set detection rate in FSOSR.

POEM: Out-of-Distribution Detection with Posterior Sampling

Out-of-distribution (OOD) detection is indispensable for machine learning models deployed in the open world. Recently, the use of an auxiliary outlier dataset during training (also known as outlier exposure) has shown promising performance. As the sample space for potential OOD data can be prohibitively large, sampling informative outliers is essential. In this work, we propose a novel posterior sampling-based outlier mining framework, POEM, which facilitates efficient use of outlier data and promotes learning a compact decision boundary between ID and OOD data for improved detection. We show that POEM establishes state-of-the-art performance on common benchmarks. Compared to the current best method that uses a greedy sampling strategy, POEM improves the relative performance by 42.0% and 24.2% (FPR95) on CIFAR-10 and CIFAR-100, respectively. We further provide theoretical insights on the effectiveness of POEM for OOD detection.

Learning from human perception to improve automatic speaker verification in style-mismatched conditions

Our prior experiments show that humans and machines seem to employ different approaches to speaker discrimination, especially in the presence of speaking style variability. The experiments examined read versus conversational speech. Listeners focused on speaker-specific idiosyncrasies while "telling speakers together", and on relative distances in a shared acoustic space when "telling speakers apart". However, automatic speaker verification (ASV) systems use the same loss function irrespective of target or non-target trials. To improve ASV performance in the presence of style variability, insights learnt from human perception are used to design a new training loss function that we refer to as "CllrCE loss". CllrCE loss uses both speaker-specific idiosyncrasies and relative acoustic distances between speakers to train the ASV system. When using the UCLA speaker variability database, in the x-vector and conditioning setups, CllrCE loss results in significant relative improvements in EER by 1-66%, and minDCF by 1-31% and 1-56%, respectively, when compared to the x-vector baseline. Using the SITW evaluation tasks, which involve different conversational speech tasks, the proposed loss combined with self-attention conditioning results in significant relative improvements in EER by 2-5% and minDCF by 6-12% over baseline. In the SITW case, performance improvements were consistent only with conditioning.

Attention-based conditioning methods using variable frame rate for style-robust speaker verification

We propose an approach to extract speaker embeddings that are robust to speaking style variations in text-independent speaker verification. Typically, speaker embedding extraction includes training a DNN for speaker classification and using the bottleneck features as speaker representations. Such a network has a pooling layer to transform frame-level to utterance-level features by calculating statistics over all utterance frames, with equal weighting. However, self-attentive embeddings perform weighted pooling such that the weights correspond to the importance of the frames in a speaker classification task. Entropy can capture acoustic variability due to speaking style variations. Hence, an entropy-based variable frame rate vector is proposed as an external conditioning vector for the self-attention layer to provide the network with information that can address style effects. This work explores five different approaches to conditioning. The best conditioning approach, concatenation with gating, provided statistically significant improvements over the x-vector baseline in 12/23 tasks and was the same as the baseline in 11/23 tasks when using the UCLA speaker variability database. It also significantly outperformed self-attention without conditioning in 9/23 tasks and was worse in 1/23. The method also showed significant improvements in multi-speaker scenarios of SITW.

TTS-CGAN: A Transformer Time-Series Conditional GAN for Biosignal Data Augmentation

Signal measurement appearing in the form of time series is one of the most common types of data used in medical machine learning applications. Such datasets are often small in size, expensive to collect and annotate, and might involve privacy issues, which hinders our ability to train large, state-of-the-art deep learning models for biomedical applications. For time-series data, the suite of data augmentation strategies we can use to expand the size of the dataset is limited by the need to maintain the basic properties of the signal. Generative Adversarial Networks (GANs) can be utilized as another data augmentation tool. In this paper, we present TTS-CGAN, a transformer-based conditional GAN model that can be trained on existing multi-class datasets and generate class-specific synthetic time-series sequences of arbitrary length. We elaborate on the model architecture and design strategies. Synthetic sequences generated by our model are indistinguishable from real ones, and can be used to complement or replace real signals of the same type, thus achieving the goal of data augmentation. To evaluate the quality of the generated data, we modify the wavelet coherence metric to be able to compare the similarity between two sets of signals, and also conduct a case study where a mix of synthetic and real data are used to train a deep learning model for sequence classification. Together with other visualization techniques and qualitative evaluation approaches, we demonstrate that TTS-CGAN generated synthetic data are similar to real data, and that our model performs better than the other state-of-the-art GAN models built for time-series data generation.

Studying Generalization Through Data Averaging

The generalization of machine learning models has a complex dependence on the data, model and learning algorithm. We study train and test performance, as well as the generalization gap given by the mean of their difference over different data set samples to understand their ``typical" behavior. We derive an expression for the gap as a function of the covariance between the model parameter distribution and the train loss, and another expression for the average test performance, showing test generalization only depends on data-averaged parameter distribution and the data-averaged loss. We show that for a large class of model parameter distributions a modified generalization gap is always non-negative. By specializing further to parameter distributions produced by stochastic gradient descent (SGD), along with a few approximations and modeling considerations, we are able to predict some aspects about how the generalization gap and model train and test performance vary as a function of SGD noise. We evaluate these predictions empirically on the Cifar10 classification task based on a ResNet architecture.

Deployment of ML Models using Kubeflow on Different Cloud Providers

This project aims to explore the process of deploying Machine learning models on Kubernetes using an open-source tool called Kubeflow [1] - an end-to-end ML Stack orchestration toolkit. We create end-to-end Machine Learning models on Kubeflow in the form of pipelines and analyze various points including the ease of setup, deployment models, performance, limitations and features of the tool. We hope that our project acts almost like a seminar/introductory report that can help vanilla cloud/Kubernetes users with zero knowledge on Kubeflow use Kubeflow to deploy ML models. From setup on different clouds to serving our trained model over the internet - we give details and metrics detailing the performance of Kubeflow.

Supervised Learning with General Risk Functionals

Standard uniform convergence results bound the generalization gap of the expected loss over a hypothesis class. The emergence of risk-sensitive learning requires generalization guarantees for functionals of the loss distribution beyond the expectation. While prior works specialize in uniform convergence of particular functionals, our work provides uniform convergence for a general class of H\"older risk functionals for which the closeness in the Cumulative Distribution Function (CDF) entails closeness in risk. We establish the first uniform convergence results for estimating the CDF of the loss distribution, yielding guarantees that hold simultaneously both over all H\"older risk functionals and over all hypotheses. Thus licensed to perform empirical risk minimization, we develop practical gradient-based methods for minimizing distortion risks (widely studied subset of H\"older risks that subsumes the spectral risks, including the mean, conditional value at risk, cumulative prospect theory risks, and others) and provide convergence guarantees. In experiments, we demonstrate the efficacy of our learning procedure, both in settings where uniform convergence results hold and in high-dimensional settings with deep networks.

On bounds for norms of reparameterized ReLU artificial neural network parameters: sums of fractional powers of the Lipschitz norm control the network parameter vector

It is an elementary fact in the scientific literature that the Lipschitz norm of the realization function of a feedforward fully-connected rectified linear unit (ReLU) artificial neural network (ANN) can, up to a multiplicative constant, be bounded from above by sums of powers of the norm of the ANN parameter vector. Roughly speaking, in this work we reveal in the case of shallow ANNs that the converse inequality is also true. More formally, we prove that the norm of the equivalence class of ANN parameter vectors with the same realization function is, up to a multiplicative constant, bounded from above by the sum of powers of the Lipschitz norm of the ANN realization function (with the exponents $ 1/2 $ and $ 1 $). Moreover, we prove that this upper bound only holds when employing the Lipschitz norm but does neither hold for H\"older norms nor for Sobolev-Slobodeckij norms. Furthermore, we prove that this upper bound only holds for sums of powers of the Lipschitz norm with the exponents $ 1/2 $ and $ 1 $ but does not hold for the Lipschitz norm alone.

Utility Theory for Sequential Decision Making

The von Neumann-Morgenstern (VNM) utility theorem shows that under certain axioms of rationality, decision-making is reduced to maximizing the expectation of some utility function. We extend these axioms to increasingly structured sequential decision making settings and identify the structure of the corresponding utility functions. In particular, we show that memoryless preferences lead to a utility in the form of a per transition reward and multiplicative factor on the future return. This result motivates a generalization of Markov Decision Processes (MDPs) with this structure on the agent's returns, which we call Affine-Reward MDPs. A stronger constraint on preferences is needed to recover the commonly used cumulative sum of scalar rewards in MDPs. A yet stronger constraint simplifies the utility function for goal-seeking agents in the form of a difference in some function of states that we call potential functions. Our necessary and sufficient conditions demystify the reward hypothesis that underlies the design of rational agents in reinforcement learning by adding an axiom to the VNM rationality axioms and motivates new directions for AI research involving sequential decision making.

Toward an ImageNet Library of Functions for Global Optimization Benchmarking

Knowledge of search-landscape features of BlackBox Optimization (BBO) problems offers valuable information in light of the Algorithm Selection and/or Configuration problems. Exploratory Landscape Analysis (ELA) models have gained success in identifying predefined human-derived features and in facilitating portfolio selectors to address those challenges. Unlike ELA approaches, the current study proposes to transform the identification problem into an image recognition problem, with a potential to detect conception-free, machine-driven landscape features. To this end, we introduce the notion of Landscape Images, which enables us to generate imagery instances per a benchmark function, and then target the classification challenge over a diverse generalized dataset of functions. We address it as a supervised multi-class image recognition problem and apply basic artificial neural network models to solve it. The efficacy of our approach is numerically validated on the noise free BBOB and IOHprofiler benchmarking suites. This evident successful learning is another step toward automated feature extraction and local structure deduction of BBO problems. By using this definition of landscape images, and by capitalizing on existing capabilities of image recognition algorithms, we foresee the construction of an ImageNet-like library of functions for training generalized detectors that rely on machine-driven features.

Nonparametric, Nonasymptotic Confidence Bands with Paley-Wiener Kernels for Band-Limited Functions

The paper introduces a method to construct confidence bands for bounded, band-limited functions based on a finite sample of input-output pairs. The approach is distribution-free w.r.t. the observation noises and only the knowledge of the input distribution is assumed. It is nonparametric, that is, it does not require a parametric model of the regression function and the regions have non-asymptotic guarantees. The algorithm is based on the theory of Paley-Wiener reproducing kernel Hilbert spaces. The paper first studies the fully observable variant, when there are no noises on the observations and only the inputs are random; then it generalizes the ideas to the noisy case using gradient-perturbation methods. Finally, numerical experiments demonstrating both cases are presented.

Patch Selection for Melanoma Classification

In medical image processing, the most important information is often located on small parts of the image. Patch-based approaches aim at using only the most relevant parts of the image. Finding ways to automatically select the patches is a challenge. In this paper, we investigate two criteria to choose patches: entropy and a spectral similarity criterion. We perform experiments at different levels of patch size. We train a Convolutional Neural Network on the subsets of patches and analyze the training time. We find that, in addition to requiring less preprocessing time, the classifiers trained on the datasets of patches selected based on entropy converge faster than on those selected based on the spectral similarity criterion and, furthermore, lead to higher accuracy. Moreover, patches of high entropy lead to faster convergence and better accuracy than patches of low entropy.

Quantification of Deep Neural Network Prediction Uncertainties for VVUQ of Machine Learning Models

Recent performance breakthroughs in Artificial intelligence (AI) and Machine learning (ML), especially advances in Deep learning (DL), the availability of powerful, easy-to-use ML libraries (e.g., scikit-learn, TensorFlow, PyTorch.), and increasing computational power have led to unprecedented interest in AI/ML among nuclear engineers. For physics-based computational models, Verification, Validation and Uncertainty Quantification (VVUQ) have been very widely investigated and a lot of methodologies have been developed. However, VVUQ of ML models has been relatively less studied, especially in nuclear engineering. In this work, we focus on UQ of ML models as a preliminary step of ML VVUQ, more specifically, Deep Neural Networks (DNNs) because they are the most widely used supervised ML algorithm for both regression and classification tasks. This work aims at quantifying the prediction, or approximation uncertainties of DNNs when they are used as surrogate models for expensive physical models. Three techniques for UQ of DNNs are compared, namely Monte Carlo Dropout (MCD), Deep Ensembles (DE) and Bayesian Neural Networks (BNNs). Two nuclear engineering examples are used to benchmark these methods, (1) time-dependent fission gas release data using the Bison code, and (2) void fraction simulation based on the BFBT benchmark using the TRACE code. It was found that the three methods typically require different DNN architectures and hyperparameters to optimize their performance. The UQ results also depend on the amount of training data available and the nature of the data. Overall, all these three methods can provide reasonable estimations of the approximation uncertainties. The uncertainties are generally smaller when the mean predictions are close to the test data, while the BNN methods usually produce larger uncertainties than MCD and DE.

Learning Controllable 3D Level Generators

Procedural Content Generation via Reinforcement Learning (PCGRL) foregoes the need for large human-authored data-sets and allows agents to train explicitly on functional constraints, using computable, user-defined measures of quality instead of target output. We explore the application of PCGRL to 3D domains, in which content-generation tasks naturally have greater complexity and potential pertinence to real-world applications. Here, we introduce several PCGRL tasks for the 3D domain, Minecraft (Mojang Studios, 2009). These tasks will challenge RL-based generators using affordances often found in 3D environments, such as jumping, multiple dimensional movement, and gravity. We train an agent to optimize each of these tasks to explore the capabilities of previous research in PCGRL. This agent is able to generate relatively complex and diverse levels, and generalize to random initial states and control targets. Controllability tests in the presented tasks demonstrate their utility to analyze success and failure for 3D generators.

Measuring and Clustering Network Attackers using Medium-Interaction Honeypots

Network honeypots are often used by information security teams to measure the threat landscape in order to secure their networks. With the advancement of honeypot development, today's medium-interaction honeypots provide a way for security teams and researchers to deploy these active defense tools that require little maintenance on a variety of protocols. In this work, we deploy such honeypots on five different protocols on the public Internet and study the intent and sophistication of the attacks we observe. We then use the information gained to develop a clustering approach that identifies correlations in attacker behavior to discover IPs that are highly likely to be controlled by a single operator, illustrating the advantage of using these honeypots for data collection.

Improved Text Classification via Test-Time Augmentation

Test-time augmentation -- the aggregation of predictions across transformed examples of test inputs -- is an established technique to improve the performance of image classification models. Importantly, TTA can be used to improve model performance post-hoc, without additional training. Although test-time augmentation (TTA) can be applied to any data modality, it has seen limited adoption in NLP due in part to the difficulty of identifying label-preserving transformations. In this paper, we present augmentation policies that yield significant accuracy improvements with language models. A key finding is that augmentation policy design -- for instance, the number of samples generated from a single, non-deterministic augmentation -- has a considerable impact on the benefit of TTA. Experiments across a binary classification task and dataset show that test-time augmentation can deliver consistent improvements over current state-of-the-art approaches.

Online Resource Allocation under Horizon Uncertainty

We study stochastic online resource allocation: a decision maker needs to allocate limited resources to stochastically-generated sequentially-arriving requests in order to maximize reward. Motivated by practice, we consider a data-driven setting in which requests are drawn independently from a distribution that is unknown to the decision maker. Online resource allocation and its special cases have been studied extensively in the past, but these previous results crucially and universally rely on a practically-untenable assumption: the total number of requests (the horizon) is known to the decision maker in advance. In many applications, such as revenue management and online advertising, the number of requests can vary widely because of fluctuations in demand or user traffic intensity. In this work, we develop online algorithms that are robust to horizon uncertainty. In sharp contrast to the known-horizon setting, we show that no algorithm can achieve a constant asymptotic competitive ratio that is independent of the horizon uncertainty. We then introduce a novel algorithm that combines dual mirror descent with a carefully-chosen target consumption sequence and prove that it achieves a bounded competitive ratio. Our algorithm is near-optimal in the sense that its competitive ratio attains the optimal rate of growth when the horizon uncertainty grows large.

On-device Synaptic Memory Consolidation using Fowler-Nordheim Quantum-tunneling

Synaptic memory consolidation has been heralded as one of the key mechanisms for supporting continual learning in neuromorphic Artificial Intelligence (AI) systems. Here we report that a Fowler-Nordheim (FN) quantum-tunneling device can implement synaptic memory consolidation similar to what can be achieved by algorithmic consolidation models like the cascade and the elastic weight consolidation (EWC) models. The proposed FN-synapse not only stores the synaptic weight but also stores the synapse's historical usage statistic on the device itself. We also show that the operation of the FN-synapse is near-optimal in terms of the synaptic lifetime and we demonstrate that a network comprising FN-synapses outperforms a comparable EWC network for a small benchmark continual learning task. With an energy footprint of femtojoules per synaptic update, we believe that the proposed FN-synapse provides an ultra-energy-efficient approach for implementing both synaptic memory consolidation and persistent learning.

BeamsNet: A data-driven Approach Enhancing Doppler Velocity Log Measurements for Autonomous Underwater Vehicle Navigation

Autonomous underwater vehicles (AUV) perform various applications such as seafloor mapping and underwater structure health monitoring. Commonly, an inertial navigation system aided by a Doppler velocity log (DVL) is used to provide the vehicle's navigation solution. In such fusion, the DVL provides the velocity vector of the AUV, which determines the navigation solution's accuracy and helps estimate the navigation states. This paper proposes BeamsNet, an end-to-end deep learning framework to regress the estimated DVL velocity vector that improves the accuracy of the velocity vector estimate, and could replace the model-based approach. Two versions of BeamsNet, differing in their input to the network, are suggested. The first uses the current DVL beam measurements and inertial sensors data, while the other utilizes only DVL data, taking the current and past DVL measurements for the regression process. Both simulation and sea experiments were made to validate the proposed learning approach relative to the model-based approach. Sea experiments were made with the Snapir AUV in the Mediterranean Sea, collecting approximately four hours of DVL and inertial sensor data. Our results show that the proposed approach achieved an improvement of more than 60% in estimating the DVL velocity vector.

Molecular Geometry Pretraining with SE(3)-Invariant Denoising Distance Matching

Pretraining molecular representations is critical in a variety of applications in drug and material discovery due to the limited number of labeled molecules, yet most of existing work focuses on pretraining on 2D molecular graphs. The power of pretraining on 3D geometric structures, however, has been less explored, owning to the difficulty of finding a sufficient proxy task to empower the pretraining to effectively extract essential features from the geometric structures. Motivated by the dynamic nature of 3D molecules, where the continuous motion of a molecule in the 3D Euclidean space forms a smooth potential energy surface, we propose a 3D coordinate denoising pretraining framework to model such an energy landscape. Leveraging a SE(3)-invariant score matching method, we propose SE(3)-DDM where the coordinate denoising proxy task is effectively boiled down to the denoising of the pairwise atomic distances in a molecule. Our comprehensive experiments confirm the effectiveness and robustness of our proposed method.

Efficient Deep Learning Using Non-Volatile Memory Technology

Embedded machine learning (ML) systems have now become the dominant platform for deploying ML serving tasks and are projected to become of equal importance for training ML models. With this comes the challenge of overall efficient deployment, in particular low power and high throughput implementations, under stringent memory constraints. In this context, non-volatile memory (NVM) technologies such as STT-MRAM and SOT-MRAM have significant advantages compared to conventional SRAM due to their non-volatility, higher cell density, and scalability features. While prior work has investigated several architectural implications of NVM for generic applications, in this work we present DeepNVM++, a comprehensive framework to characterize, model, and analyze NVM-based caches in GPU architectures for deep learning (DL) applications by combining technology-specific circuit-level models and the actual memory behavior of various DL workloads. DeepNVM++ relies on iso-capacity and iso-area performance and energy models for last-level caches implemented using conventional SRAM and emerging STT-MRAM and SOT-MRAM technologies. In the iso-capacity case, STT-MRAM and SOT-MRAM provide up to 3.8x and 4.7x energy-delay product (EDP) reduction and 2.4x and 2.8x area reduction compared to conventional SRAM, respectively. Under iso-area assumptions, STT-MRAM and SOT-MRAM provide up to 2.2x and 2.4x EDP reduction and accommodate 2.3x and 3.3x cache capacity when compared to SRAM, respectively. We also perform a scalability analysis and show that STT-MRAM and SOT-MRAM achieve orders of magnitude EDP reduction when compared to SRAM for large cache capacities. DeepNVM++ is demonstrated on STT-/SOT-MRAM technologies and can be used for the characterization, modeling, and analysis of any NVM technology for last-level caches in GPUs for DL applications.

Challenges and Opportunities in Multi-device Speech Processing

We review current solutions and technical challenges for automatic speech recognition, keyword spotting, device arbitration, speech enhancement, and source localization in multidevice home environments to provide context for the INTERSPEECH 2022 special session, "Challenges and opportunities for signal processing and machine learning for multiple smart devices". We also identify the datasets needed to support these research areas. Based on the review and our research experience in the multi-device domain, we conclude with an outlook on the future evolution

Reduced Optimal Power Flow Using Graph Neural Network

OPF problems are formulated and solved for power system operations, especially for determining generation dispatch points in real-time. For large and complex power system networks with large numbers of variables and constraints, finding the optimal solution for real-time OPF in a timely manner requires a massive amount of computing power. This paper presents a new method to reduce the number of constraints in the original OPF problem using a graph neural network (GNN). GNN is an innovative machine learning model that utilizes features from nodes, edges, and network topology to maximize its performance. In this paper, we proposed a GNN model to predict which lines would be heavily loaded or congested with given load profiles and generation capacities. Only these critical lines will be monitored in an OPF problem, creating a reduced OPF (ROPF) problem. Significant saving in computing time is expected from the proposed ROPF model. A comprehensive analysis of predictions from the GNN model was also made. It is concluded that the application of GNN for ROPF is able to reduce computing time while retaining solution quality.

Heterogeneous mixtures of dictionary functions to approximate subspace invariance in Koopman operators

Koopman operators model nonlinear dynamics as a linear dynamic system acting on a nonlinear function as the state. This nonstandard state is often called a Koopman observable and is usually approximated numerically by a superposition of functions drawn from a \textit{dictionary}. A widely used algorithm, is \textit{Extended Dynamic Mode Decomposition}, where the dictionary functions are drawn from a fixed, homogeneous class of functions. Recently, deep learning combined with EDMD has been used to learn novel dictionary functions in an algorithm called deep dynamic mode decomposition (deepDMD). The learned representation both (1) accurately models and (2) scales well with the dimension of the original nonlinear system. In this paper we analyze the learned dictionaries from deepDMD and explore the theoretical basis for their strong performance. We discover a novel class of dictionary functions to approximate Koopman observables. Error analysis of these dictionary functions show they satisfy a property of subspace approximation, which we define as uniform finite approximate closure. We discover that structured mixing of heterogeneous dictionary functions drawn from different classes of nonlinear functions achieve the same accuracy and dimensional scaling as deepDMD. This mixed dictionary does so with an order of magnitude reduction in parameters, while maintaining geometric interpretability. Our results provide a hypothesis to explain the success of deep neural networks in learning numerical approximations to Koopman operators.

Exact Spectral Norm Regularization for Neural Networks

We pursue a line of research that seeks to regularize the spectral norm of the Jacobian of the input-output mapping for deep neural networks. While previous work rely on upper bounding techniques, we provide a scheme that targets the exact spectral norm. We showcase that our algorithm achieves an improved generalization performance compared to previous spectral regularization techniques while simultaneously maintaining a strong safeguard against natural and adversarial noise. Moreover, we further explore some previous reasoning concerning the strong adversarial protection that Jacobian regularization provides and show that it can be misleading.

Rankings from multimodal pairwise comparisons

The task of ranking individuals or teams, based on a set of comparisons between pairs, arises in various contexts, including sporting competitions and the analysis of dominance hierarchies among animals and humans. Given data on which competitors beat which others, the challenge is to rank the competitors from best to worst. Here we study the problem of computing rankings when there are multiple, potentially conflicting modes of comparison, such as multiple types of dominance behaviors among animals. We assume that we do not know a priori what information each behavior conveys about the ranking, or even whether they convey any information at all. Nonetheless we show that it is possible to compute a ranking in this situation and present a fast method for doing so, based on a combination of an expectation-maximization algorithm and a modified Bradley-Terry model. We give a selection of example applications to both animal and human competition.

Materials Transformers Language Models for Generative Materials Design: a benchmark study

Pre-trained transformer language models on large unlabeled corpus have produced state-of-the-art results in natural language processing, organic molecule design, and protein sequence generation. However, no such models have been applied to learn the composition patterns of inorganic materials. Here we train a series of seven modern transformer language models (GPT, GPT-2, GPT-Neo, GPT-J, BLMM, BART, and RoBERTa) using the expanded formulas from material deposited in the ICSD, OQMD, and Materials Projects databases. Six different datasets with/out non-charge-neutral or balanced electronegativity samples are used to benchmark the performances and uncover the generation biases of modern transformer models for the generative design of materials compositions. Our extensive experiments showed that the causal language models based materials transformers can generate chemically valid materials compositions with as high as 97.54\% to be charge neutral and 91.40\% to be electronegativity balanced, which has more than 6 times higher enrichment compared to a baseline pseudo-random sampling algorithm. These models also demonstrate high novelty and their potential in new materials discovery has been proved by their capability to recover the leave-out materials. We also find that the properties of the generated samples can be tailored by training the models with selected training sets such as high-bandgap materials. Our experiments also showed that different models each have their own preference in terms of the properties of the generated samples and their running time complexity varies a lot. We have applied our materials transformer models to discover a set of new materials as validated using DFT calculations.

A View Independent Classification Framework for Yoga Postures

Yoga is a globally acclaimed and widely recommended practice for a healthy living. Maintaining correct posture while performing a Yogasana is of utmost importance. In this work, we employ transfer learning from Human Pose Estimation models for extracting 136 key-points spread all over the body to train a Random Forest classifier which is used for estimation of the Yogasanas. The results are evaluated on an in-house collected extensive yoga video database of 51 subjects recorded from 4 different camera angles. We propose a 3 step scheme for evaluating the generalizability of a Yoga classifier by testing it on 1) unseen frames, 2) unseen subjects, and 3) unseen camera angles. We argue that for most of the applications, validation accuracies on unseen subjects and unseen camera angles would be most important. We empirically analyze over three public datasets, the advantage of transfer learning and the possibilities of target leakage. We further demonstrate that the classification accuracies critically depend on the cross validation method employed and can often be misleading. To promote further research, we have made key-points dataset and code publicly available.

AutoInit: Automatic Initialization via Jacobian Tuning

Good initialization is essential for training Deep Neural Networks (DNNs). Oftentimes such initialization is found through a trial and error approach, which has to be applied anew every time an architecture is substantially modified, or inherited from smaller size networks leading to sub-optimal initialization. In this work we introduce a new and cheap algorithm, that allows one to find a good initialization automatically, for general feed-forward DNNs. The algorithm utilizes the Jacobian between adjacent network blocks to tune the network hyperparameters to criticality. We solve the dynamics of the algorithm for fully connected networks with ReLU and derive conditions for its convergence. We then extend the discussion to more general architectures with BatchNorm and residual connections. Finally, we apply our method to ResMLP and VGG architectures, where the automatic one-shot initialization found by our method shows good performance on vision tasks.

Programmatic Concept Learning for Human Motion Description and Synthesis

We introduce Programmatic Motion Concepts, a hierarchical motion representation for human actions that captures both low-level motion and high-level description as motion concepts. This representation enables human motion description, interactive editing, and controlled synthesis of novel video sequences within a single framework. We present an architecture that learns this concept representation from paired video and action sequences in a semi-supervised manner. The compactness of our representation also allows us to present a low-resource training recipe for data-efficient learning. By outperforming established baselines, especially in the small data regime, we demonstrate the efficiency and effectiveness of our framework for multiple applications.

Neural Neural Textures Make Sim2Real Consistent

Unpaired image translation algorithms can be used for sim2real tasks, but many fail to generate temporally consistent results. We present a new approach that combines differentiable rendering with image translation to achieve temporal consistency over indefinite timescales, using surface consistency losses and \emph{neural neural textures}. We call this algorithm TRITON (Texture Recovering Image Translation Network): an unsupervised, end-to-end, stateless sim2real algorithm that leverages the underlying 3D geometry of input scenes by generating realistic-looking learnable neural textures. By settling on a particular texture for the objects in a scene, we ensure consistency between frames statelessly. Unlike previous algorithms, TRITON is not limited to camera movements -- it can handle the movement of objects as well, making it useful for downstream tasks such as robotic manipulation.

Prompting Decision Transformer for Few-Shot Policy Generalization

Humans can leverage prior experience and learn novel tasks from a handful of demonstrations. In contrast to offline meta-reinforcement learning, which aims to achieve quick adaptation through better algorithm design, we investigate the effect of architecture inductive bias on the few-shot learning capability. We propose a Prompt-based Decision Transformer (Prompt-DT), which leverages the sequential modeling ability of the Transformer architecture and the prompt framework to achieve few-shot adaptation in offline RL. We design the trajectory prompt, which contains segments of the few-shot demonstrations, and encodes task-specific information to guide policy generation. Our experiments in five MuJoCo control benchmarks show that Prompt-DT is a strong few-shot learner without any extra finetuning on unseen target tasks. Prompt-DT outperforms its variants and strong meta offline RL baselines by a large margin with a trajectory prompt containing only a few timesteps. Prompt-DT is also robust to prompt length changes and can generalize to out-of-distribution (OOD) environments.

Auditing Visualizations: Transparency Methods Struggle to Detect Anomalous Behavior

Transparency methods such as model visualizations provide information that outputs alone might miss, since they describe the internals of neural networks. But can we trust that model explanations reflect model behavior? For instance, can they diagnose abnormal behavior such as backdoors or shape bias? To evaluate model explanations, we define a model as anomalous if it differs from a reference set of normal models, and we test whether transparency methods assign different explanations to anomalous and normal models. We find that while existing methods can detect stark anomalies such as shape bias or adversarial training, they struggle to identify more subtle anomalies such as models trained on incomplete data. Moreover, they generally fail to distinguish the inputs that induce anomalous behavior, e.g. images containing a backdoor trigger. These results reveal new blind spots in existing model explanations, pointing to the need for further method development.

Robustness Implies Generalization via Data-Dependent Generalization Bounds

This paper proves that robustness implies generalization via data-dependent generalization bounds. As a result, robustness and generalization are shown to be connected closely in a data-dependent manner. Our bounds improve previous bounds in two directions, to solve an open problem that has seen little development since 2010. The first is to reduce the dependence on the covering number. The second is to remove the dependence on the hypothesis space. We present several examples, including ones for lasso and deep learning, in which our bounds are provably preferable. The experiments on real-world data and theoretical models demonstrate near-exponential improvements in various situations. To achieve these improvements, we do not require additional assumptions on the unknown distribution; instead, we only incorporate an observable and computable property of the training samples. A key technical innovation is an improved concentration bound for multinomial random variables that is of independent interest beyond robustness and generalization.

ProGen2: Exploring the Boundaries of Protein Language Models

Attention-based models trained on protein sequences have demonstrated incredible success at classification and generation tasks relevant for artificial intelligence-driven protein design. However, we lack a sufficient understanding of how very large-scale models and data play a role in effective protein model development. We introduce a suite of protein language models, named ProGen2, that are scaled up to 6.4B parameters and trained on different sequence datasets drawn from over a billion proteins from genomic, metagenomic, and immune repertoire databases. ProGen2 models show state-of-the-art performance in capturing the distribution of observed evolutionary sequences, generating novel viable sequences, and predicting protein fitness without additional finetuning. As large model sizes and raw numbers of protein sequences continue to become more widely accessible, our results suggest that a growing emphasis needs to be placed on the data distribution provided to a protein sequence model. We release the ProGen2 models and code at https://github.com/salesforce/progen.

Effective training-time stacking for ensembling of deep neural networks

Ensembling is a popular and effective method for improving machine learning (ML) models. It proves its value not only in classical ML but also for deep learning. Ensembles enhance the quality and trustworthiness of ML solutions, and allow uncertainty estimation. However, they come at a price: training ensembles of deep learning models eat a huge amount of computational resources. A snapshot ensembling collects models in the ensemble along a single training path. As it runs training only one time, the computational time is similar to the training of one model. However, the quality of models along the training path is different: typically, later models are better if no overfitting occurs. So, the models are of varying utility. Our method improves snapshot ensembling by selecting and weighting ensemble members along the training path. It relies on training-time likelihoods without looking at validation sample errors that standard stacking methods do. Experimental evidence for Fashion MNIST, CIFAR-10, and CIFAR-100 datasets demonstrates the superior quality of the proposed weighted ensembles c.t. vanilla ensembling of deep learning models.

Supply-Side Equilibria in Recommender Systems

Digital recommender systems such as Spotify and Netflix affect not only consumer behavior but also producer incentives: producers seek to supply content that will be recommended by the system. But what content will be produced? In this paper, we investigate the supply-side equilibria in content recommender systems. We model users and content as $D$-dimensional vectors, and recommend the content that has the highest dot product with each user. The main features of our model are that the producer decision space is high-dimensional and the user base is heterogeneous. This gives rise to new qualitative phenomena at equilibrium: First, the formation of genres, where producers specialize to compete for subsets of users. Using a duality argument, we derive necessary and sufficient conditions for this specialization to occur. Second, we show that producers can achieve positive profit at equilibrium, which is typically impossible under perfect competition. We derive sufficient conditions for this to occur, and show it is closely connected to specialization of content. In both results, the interplay between the geometry of the users and the structure of producer costs influences the structure of the supply-side equilibria. At a conceptual level, our work serves as a starting point to investigate how recommender systems shape supply-side competition between producers.

Positive-definite parametrization of mixed quantum states with deep neural networks

We introduce the Gram-Hadamard Density Operator (GHDO), a new deep neural-network architecture that can encode positive semi-definite density operators of exponential rank with polynomial resources. We then show how to embed an autoregressive structure in the GHDO to allow direct sampling of the probability distribution. These properties are especially important when representing and variationally optimizing the mixed quantum state of a system interacting with an environment. Finally, we benchmark this architecture by simulating the steady state of the dissipative transverse-field Ising model. Estimating local observables and the R\'enyi entropy, we show significant improvements over previous state-of-the-art variational approaches.

Understanding Benign Overfitting in Nested Meta Learning

Meta learning has demonstrated tremendous success in few-shot learning with limited supervised data. In those settings, the meta model is usually overparameterized. While the conventional statistical learning theory suggests that overparameterized models tend to overfit, empirical evidence reveals that overparameterized meta learning methods still work well -- a phenomenon often called ``benign overfitting.'' To understand this phenomenon, we focus on the meta learning settings with a challenging nested structure that we term the nested meta learning, and analyze its generalization performance under an overparameterized meta learning model. While our analysis uses the relatively tractable linear models, our theory contributes to understanding the delicate interplay among data heterogeneity, model adaptation and benign overfitting in nested meta learning tasks. We corroborate our theoretical claims through numerical simulations.

Impact of Acoustic Event Tagging on Scene Classification in a Multi-Task Learning Framework

Acoustic events are sounds with well-defined spectro-temporal characteristics which can be associated with the physical objects generating them. Acoustic scenes are collections of such acoustic events in no specific temporal order. Given this natural linkage between events and scenes, a common belief is that the ability to classify events must help in the classification of scenes. This has led to several efforts attempting to do well on Acoustic Event Tagging (AET) and Acoustic Scene Classification (ASC) using a multi-task network. However, in these efforts, improvement in one task does not guarantee an improvement in the other, suggesting a tension between ASC and AET. It is unclear if improvements in AET translates to improvements in ASC. We explore this conundrum through an extensive empirical study and show that under certain conditions, using AET as an auxiliary task in the multi-task network consistently improves ASC performance. Additionally, ASC performance further improves with the AET data-set size and is not sensitive to the choice of events or the number of events in the AET data-set. We conclude that this improvement in ASC performance comes from the regularization effect of using AET and not from the network's improved ability to discern between acoustic events.

Thermodynamics of Interpretation

Over the past few years, different types of data-driven Artificial Intelligence (AI) techniques have been widely adopted in various domains of science for generating predictive black-box models. However, because of their black-box nature, it is crucial to establish trust in these models before accepting them as accurate. One way of achieving this goal is through the implementation of a post-hoc interpretation scheme that can put forward the reasons behind a black-box model prediction. In this work, we propose a classical thermodynamics inspired approach for this purpose: Thermodynamically Explainable Representations of AI and other black-box Paradigms (TERP). TERP works by constructing a linear, local surrogate model that approximates the behaviour of the black-box model within a small neighborhood around the instance being explained. By employing a simple forward feature selection Monte Carlo algorithm, TERP assigns an interpretability free energy score to all the possible surrogate models in order to choose an optimal interpretation. Additionally, we validate TERP as a generally applicable method by successfully interpreting four different classes of black-box models trained on datasets coming from relevant domains, including classifying images, predicting heart disease and classifying biomolecular conformations.

Variational Autoencoder Assisted Neural Network Likelihood RSRP Prediction Model

Measuring customer experience on mobile data is of utmost importance for global mobile operators. The reference signal received power (RSRP) is one of the important indicators for current mobile network management, evaluation and monitoring. Radio data gathered through the minimization of drive test (MDT), a 3GPP standard technique, is commonly used for radio network analysis. Collecting MDT data in different geographical areas is inefficient and constrained by the terrain conditions and user presence, hence is not an adequate technique for dynamic radio environments. In this paper, we study a generative model for RSRP prediction, exploiting MDT data and a digital twin (DT), and propose a data-driven, two-tier neural network (NN) model. In the first tier, environmental information related to user equipment (UE), base stations (BS) and network key performance indicators (KPI) are extracted through a variational autoencoder (VAE). The second tier is designed as a likelihood model. Here, the environmental features and real MDT data features are adopted, formulating an integrated training process. On validation, our proposed model that uses real-world data demonstrates an accuracy improvement of about 20% or more compared with the empirical model and about 10% when compared with a fully connected prediction network.

Iso-CapsNet: Isomorphic Capsule Network for Brain Graph Representation Learning

Brain graph representation learning serves as the fundamental technique for brain diseases diagnosis. Great efforts from both the academic and industrial communities have been devoted to brain graph representation learning in recent years. The isomorphic neural network (IsoNN) introduced recently can automatically learn the existence of sub-graph patterns in brain graphs, which is also the state-of-the-art brain graph representation learning method by this context so far. However, IsoNN fails to capture the orientations of sub-graph patterns, which may render the learned representations to be useless for many cases. In this paper, we propose a new Iso-CapsNet (Isomorphic Capsule Net) model by introducing the graph isomorphic capsules for effective brain graph representation learning. Based on the capsule dynamic routing, besides the subgraph pattern existence confidence scores, Iso-CapsNet can also learn other sub-graph rich properties, including position, size and orientation, for calculating the class-wise digit capsules. We have compared Iso-CapsNet with both classic and state-of-the-art brain graph representation approaches with extensive experiments on four brain graph benchmark datasets. The experimental results also demonstrate the effectiveness of Iso-CapsNet, which can out-perform the baseline methods with significant improvements.

When to Trust Your Simulator: Dynamics-Aware Hybrid Offline-and-Online Reinforcement Learning

Learning effective reinforcement learning (RL) policies to solve real-world complex tasks can be quite challenging without a high-fidelity simulation environment. In most cases, we are only given imperfect simulators with simplified dynamics, which inevitably lead to severe sim-to-real gaps in RL policy learning. The recently emerged field of offline RL provides another possibility to learn policies directly from pre-collected historical data. However, to achieve reasonable performance, existing offline RL algorithms need impractically large offline data with sufficient state-action space coverage for training. This brings up a new question: is it possible to combine learning from limited real data in offline RL and unrestricted exploration through imperfect simulators in online RL to address the drawbacks of both approaches? In this study, we propose the Dynamics-Aware Hybrid Offline-and-Online Reinforcement Learning (H2O) framework to provide an affirmative answer to this question. H2O introduces a dynamics-aware policy evaluation scheme, which adaptively penalizes the Q function learning on simulated state-action pairs with large dynamics gaps, while also simultaneously allowing learning from a fixed real-world dataset. Through extensive simulation and real-world tasks, as well as theoretical analysis, we demonstrate the superior performance of H2O against other cross-domain online and offline RL algorithms. H2O provides a brand new hybrid offline-and-online RL paradigm, which can potentially shed light on future RL algorithm design for solving practical real-world tasks.

"Double vaccinated, 5G boosted!": Learning Attitudes towards COVID-19 Vaccination from Social Media

To address the vaccine hesitancy which impairs the efforts of the COVID-19 vaccination campaign, it is imperative to understand public vaccination attitudes and timely grasp their changes. In spite of reliability and trustworthiness, conventional attitude collection based on surveys is time-consuming and expensive, and cannot follow the fast evolution of vaccination attitudes. We leverage the textual posts on social media to extract and track users' vaccination stances in near real time by proposing a deep learning framework. To address the impact of linguistic features such as sarcasm and irony commonly used in vaccine-related discourses, we integrate into the framework the recent posts of a user's social network neighbours to help detect the user's genuine attitude. Based on our annotated dataset from Twitter, the models instantiated from our framework can increase the performance of attitude extraction by up to 23% compared to state-of-the-art text-only models. Using this framework, we successfully validate the feasibility of using social media to track the evolution of vaccination attitudes in real life. We further show one practical use of our framework by validating the possibility to forecast a user's vaccine hesitancy changes with information perceived from social media.

Causal Dynamics Learning for Task-Independent State Abstraction

Learning dynamics models accurately is an important goal for Model-Based Reinforcement Learning (MBRL), but most MBRL methods learn a dense dynamics model which is vulnerable to spurious correlations and therefore generalizes poorly to unseen states. In this paper, we introduce Causal Dynamics Learning for Task-Independent State Abstraction (CDL), which first learns a theoretically proved causal dynamics model that removes unnecessary dependencies between state variables and the action, thus generalizing well to unseen states. A state abstraction can then be derived from the learned dynamics, which not only improves sample efficiency but also applies to a wider range of tasks than existing state abstraction methods. Evaluated on two simulated environments and downstream tasks, both the dynamics model and policies learned by the proposed method generalize well to unseen states and the derived state abstraction improves sample efficiency compared to learning without it.

Distinguishing Learning Rules with Brain Machine Interfaces

Despite extensive theoretical work on biologically plausible learning rules, it has been difficult to obtain clear evidence about whether and how such rules are implemented in the brain. We consider biologically plausible supervised- and reinforcement-learning rules and ask whether changes in network activity during learning can be used to determine which learning rule is being used. Supervised learning requires a credit-assignment model estimating the mapping from neural activity to behavior, and, in a biological organism, this model will inevitably be an imperfect approximation of the ideal mapping, leading to a bias in the direction of the weight updates relative to the true gradient. Reinforcement learning, on the other hand, requires no credit-assignment model and tends to make weight updates following the true gradient direction. We derive a metric to distinguish between learning rules by observing changes in the network activity during learning, given that the mapping from brain to behavior is known by the experimenter. Because brain-machine interface (BMI) experiments allow for perfect knowledge of this mapping, we focus on modeling a cursor-control BMI task using recurrent neural networks, showing that learning rules can be distinguished in simulated experiments using only observations that a neuroscience experimenter would plausibly have access to.

Pen and Paper Exercises in Machine Learning

This is a collection of (mostly) pen-and-paper exercises in machine learning. The exercises are on the following topics: linear algebra, optimisation, directed graphical models, undirected graphical models, expressive power of graphical models, factor graphs and message passing, inference for hidden Markov models, model-based learning (including ICA and unnormalised models), sampling and Monte-Carlo integration, and variational inference.

ContraReg: Contrastive Learning of Multi-modality Unsupervised Deformable Image Registration

Establishing voxelwise semantic correspondence across distinct imaging modalities is a foundational yet formidable computer vision task. Current multi-modality registration techniques maximize hand-crafted inter-domain similarity functions, are limited in modeling nonlinear intensity-relationships and deformations, and may require significant re-engineering or underperform on new tasks, datasets, and domain pairs. This work presents ContraReg, an unsupervised contrastive representation learning approach to multi-modality deformable registration. By projecting learned multi-scale local patch features onto a jointly learned inter-domain embedding space, ContraReg obtains representations useful for non-rigid multi-modality alignment. Experimentally, ContraReg achieves accurate and robust results with smooth and invertible deformations across a series of baselines and ablations on a neonatal T1-T2 brain MRI registration task with all methods validated over a wide range of deformation regularization strengths.

Interpretable Hidden Markov Model-Based Deep Reinforcement Learning Hierarchical Framework for Predictive Maintenance of Turbofan Engines

An open research question in deep reinforcement learning is how to focus the policy learning of key decisions within a sparse domain. This paper emphasizes combining the advantages of inputoutput hidden Markov models and reinforcement learning towards interpretable maintenance decisions. We propose a novel hierarchical-modeling methodology that, at a high level, detects and interprets the root cause of a failure as well as the health degradation of the turbofan engine, while, at a low level, it provides the optimal replacement policy. It outperforms the baseline performance of deep reinforcement learning methods applied directly to the raw data or when using a hidden Markov model without such a specialized hierarchy. It also provides comparable performance to prior work, however, with the additional benefit of interpretability.

Benchopt: Reproducible, efficient and collaborative optimization benchmarks

Numerical validation is at the core of machine learning research as it allows to assess the actual impact of new methods, and to confirm the agreement between theory and practice. Yet, the rapid development of the field poses several challenges: researchers are confronted with a profusion of methods to compare, limited transparency and consensus on best practices, as well as tedious re-implementation work. As a result, validation is often very partial, which can lead to wrong conclusions that slow down the progress of research. We propose Benchopt, a collaborative framework to automate, reproduce and publish optimization benchmarks in machine learning across programming languages and hardware architectures. Benchopt simplifies benchmarking for the community by providing an off-the-shelf tool for running, sharing and extending experiments. To demonstrate its broad usability, we showcase benchmarks on three standard learning tasks: $\ell_2$-regularized logistic regression, Lasso, and ResNet18 training for image classification. These benchmarks highlight key practical findings that give a more nuanced view of the state-of-the-art for these problems, showing that for practical evaluation, the devil is in the details. We hope that Benchopt will foster collaborative work in the community hence improving the reproducibility of research findings.

DeStripe: A Self2Self Spatio-Spectral Graph Neural Network with Unfolded Hessian for Stripe Artifact Removal in Light-sheet Microscopy

Light-sheet fluorescence microscopy (LSFM) is a cutting-edge volumetric imaging technique that allows for three-dimensional imaging of mesoscopic samples with decoupled illumination and detection paths. Although the selective excitation scheme of such a microscope provides intrinsic optical sectioning that minimizes out-of-focus fluorescence background and sample photodamage, it is prone to light absorption and scattering effects, which results in uneven illumination and striping artifacts in the images adversely. To tackle this issue, in this paper, we propose a blind stripe artifact removal algorithm in LSFM, called DeStripe, which combines a self-supervised spatio-spectral graph neural network with unfolded Hessian prior. Specifically, inspired by the desirable properties of Fourier transform in condensing striping information into isolated values in the frequency domain, DeStripe firstly localizes the potentially corrupted Fourier coefficients by exploiting the structural difference between unidirectional stripe artifacts and more isotropic foreground images. Affected Fourier coefficients can then be fed into a graph neural network for recovery, with a Hessian regularization unrolled to further ensure structures in the standard image space are well preserved. Since in realistic, stripe-free LSFM barely exists with a standard image acquisition protocol, DeStripe is equipped with a Self2Self denoising loss term, enabling artifact elimination without access to stripe-free ground truth images. Competitive experimental results demonstrate the efficacy of DeStripe in recovering corrupted biomarkers in LSFM with both synthetic and real stripe artifacts.

Learning To Cut By Looking Ahead: Cutting Plane Selection via Imitation Learning

Cutting planes are essential for solving mixed-integer linear problems (MILPs), because they facilitate bound improvements on the optimal solution value. For selecting cuts, modern solvers rely on manually designed heuristics that are tuned to gauge the potential effectiveness of cuts. We show that a greedy selection rule explicitly looking ahead to select cuts that yield the best bound improvement delivers strong decisions for cut selection - but is too expensive to be deployed in practice. In response, we propose a new neural architecture (NeuralCut) for imitation learning on the lookahead expert. Our model outperforms standard baselines for cut selection on several synthetic MILP benchmarks. Experiments with a B&C solver for neural network verification further validate our approach, and exhibit the potential of learning methods in this setting.

Explicitly incorporating spatial information to recurrent networks for agriculture

In agriculture, the majority of vision systems perform still image classification. Yet, recent work has highlighted the potential of spatial and temporal cues as a rich source of information to improve the classification performance. In this paper, we propose novel approaches to explicitly capture both spatial and temporal information to improve the classification of deep convolutional neural networks. We leverage available RGB-D images and robot odometry to perform inter-frame feature map spatial registration. This information is then fused within recurrent deep learnt models, to improve their accuracy and robustness. We demonstrate that this can considerably improve the classification performance with our best performing spatial-temporal model (ST-Atte) achieving absolute performance improvements for intersection-over-union (IoU[%]) of 4.7 for crop-weed segmentation and 2.6 for fruit (sweet pepper) segmentation. Furthermore, we show that these approaches are robust to variable framerates and odometry errors, which are frequently observed in real-world applications.

Utilizing Class Separation Distance for the Evaluation of Corruption Robustness of Machine Learning Classifiers

Robustness is a fundamental pillar of Machine Learning (ML) classifiers, substantially determining their reliability. Methods for assessing classifier robustness are therefore essential. In this work, we address the challenge of evaluating corruption robustness in a way that allows comparability and interpretability on a given dataset. We propose a test data augmentation method that uses a robustness distance $\epsilon$ derived from the datasets minimal class separation distance. The resulting MSCR (mean statistical corruption robustness) metric allows a dataset-specific comparison of different classifiers with respect to their corruption robustness. The MSCR value is interpretable, as it represents the classifiers avoidable loss of accuracy due to statistical corruptions. On 2D and image data, we show that the metric reflects different levels of classifier robustness. Furthermore, we observe unexpected optima in classifiers robust accuracy through training and testing classifiers with different levels of noise. While researchers have frequently reported on a significant tradeoff on accuracy when training robust models, we strengthen the view that a tradeoff between accuracy and corruption robustness is not inherent. Our results indicate that robustness training through simple data augmentation can already slightly improve accuracy.

Transfer Learning via Test-Time Neural Networks Aggregation

It has been demonstrated that deep neural networks outperform traditional machine learning. However, deep networks lack generalisability, that is, they will not perform as good as in a new (testing) set drawn from a different distribution due to the domain shift. In order to tackle this known issue, several transfer learning approaches have been proposed, where the knowledge of a trained model is transferred into another to improve performance with different data. However, most of these approaches require additional training steps, or they suffer from catastrophic forgetting that occurs when a trained model has overwritten previously learnt knowledge. We address both problems with a novel transfer learning approach that uses network aggregation. We train dataset-specific networks together with an aggregation network in a unified framework. The loss function includes two main components: a task-specific loss (such as cross-entropy) and an aggregation loss. The proposed aggregation loss allows our model to learn how trained deep network parameters can be aggregated with an aggregation operator. We demonstrate that the proposed approach learns model aggregation at test time without any further training step, reducing the burden of transfer learning to a simple arithmetical operation. The proposed approach achieves comparable performance w.r.t. the baseline. Besides, if the aggregation operator has an inverse, we will show that our model also inherently allows for selective forgetting, i.e., the aggregated model can forget one of the datasets it was trained on, retaining information on the others.

Guillotine Regularization: Improving Deep Networks Generalization by Removing their Head

One unexpected technique that emerged in recent years consists in training a Deep Network (DN) with a Self-Supervised Learning (SSL) method, and using this network on downstream tasks but with its last few layers entirely removed. This usually skimmed-over trick is actually critical for SSL methods to display competitive performances. For example, on ImageNet classification, more than 30 points of percentage can be gained that way. This is a little vexing, as one would hope that the network layer at which invariance is explicitly enforced by the SSL criterion during training (the last layer) should be the one to use for best generalization performance downstream. But it seems not to be, and this study sheds some light on why. This trick, which we name Guillotine Regularization (GR), is in fact a generically applicable form of regularization that has also been used to improve generalization performance in transfer learning scenarios. In this work, through theory and experiments, we formalize GR and identify the underlying reasons behind its success in SSL methods. Our study shows that the use of this trick is essential to SSL performance for two main reasons: (i) improper data-augmentations to define the positive pairs used during training, and/or (ii) suboptimal selection of the hyper-parameters of the SSL loss.

Stability Verification of Neural Network Controllers using Mixed-Integer Programming

We propose a framework for the stability verification of Mixed-Integer Linear Programming (MILP) representable control policies. This framework compares a fixed candidate policy, which admits an efficient parameterization and can be evaluated at a low computational cost, against a fixed baseline policy, which is known to be stable but expensive to evaluate. We provide sufficient conditions for the closed-loop stability of the candidate policy in terms of the worst-case approximation error with respect to the baseline policy, and we show that these conditions can be checked by solving a Mixed-Integer Quadratic Program (MIQP). Additionally, we demonstrate that an outer approximation of the stability region of the candidate policy can be computed by solving an MILP. The proposed framework is sufficiently general to accommodate a broad range of candidate policies including ReLU Neural Networks (NNs), optimal solution maps of parametric quadratic programs, and Model Predictive Control (MPC) policies. We also present an open-source toolbox in Python based on the proposed framework, which allows for the easy verification of custom NN architectures and MPC formulations. We showcase the flexibility and reliability of our framework in the context of a DC-DC power convertor case study and investigate the computational complexity.

Interpretable Acoustic Representation Learning on Breathing and Speech Signals for COVID-19 Detection

In this paper, we describe an approach for representation learning of audio signals for the task of COVID-19 detection. The raw audio samples are processed with a bank of 1-D convolutional filters that are parameterized as cosine modulated Gaussian functions. The choice of these kernels allows the interpretation of the filterbanks as smooth band-pass filters. The filtered outputs are pooled, log-compressed and used in a self-attention based relevance weighting mechanism. The relevance weighting emphasizes the key regions of the time-frequency decomposition that are important for the downstream task. The subsequent layers of the model consist of a recurrent architecture and the models are trained for a COVID-19 detection task. In our experiments on the Coswara data set, we show that the proposed model achieves significant performance improvements over the baseline system as well as other representation learning approaches. Further, the approach proposed is shown to be uniformly applicable for speech and breathing signals and for transfer learning from a larger data set.

A Simple and Scalable Tensor Completion Algorithm via Latent Invariant Constraint for Recommendation System

In this paper we provide a latent-variable formulation and solution to the recommender system (RS) problem in terms of a fundamental property that any reasonable solution should be expected to satisfy. Specifically, we examine a novel tensor completion method to efficiently and accurately learn parameters of a model for the unobservable personal preferences that underly user ratings. By regularizing the tensor decomposition with a single latent invariant, we achieve three properties for a reliable recommender system: (1) uniqueness of the tensor completion result with minimal assumptions, (2) unit consistency that is independent of arbitrary preferences of users, and (3) a consensus ordering guarantee that provides consistent ranking between observed and unobserved rating scores. Our algorithm leads to a simple and elegant recommendation framework that has linear computational complexity and with no hyperparameter tuning. We provide empirical results demonstrating that the approach significantly outperforms current state-of-the-art methods.

Benign overfitting and adaptive nonparametric regression

In the nonparametric regression setting, we construct an estimator which is a continuous function interpolating the data points with high probability, while attaining minimax optimal rates under mean squared risk on the scale of H\"older classes adaptively to the unknown smoothness.

Distributional Gaussian Processes Layers for Out-of-Distribution Detection

Machine learning models deployed on medical imaging tasks must be equipped with out-of-distribution detection capabilities in order to avoid erroneous predictions. It is unsure whether out-of-distribution detection models reliant on deep neural networks are suitable for detecting domain shifts in medical imaging. Gaussian Processes can reliably separate in-distribution data points from out-of-distribution data points via their mathematical construction. Hence, we propose a parameter efficient Bayesian layer for hierarchical convolutional Gaussian Processes that incorporates Gaussian Processes operating in Wasserstein-2 space to reliably propagate uncertainty. This directly replaces convolving Gaussian Processes with a distance-preserving affine operator on distributions. Our experiments on brain tissue-segmentation show that the resulting architecture approaches the performance of well-established deterministic segmentation algorithms (U-Net), which has not been achieved with previous hierarchical Gaussian Processes. Moreover, by applying the same segmentation model to out-of-distribution data (i.e., images with pathology such as brain tumors), we show that our uncertainty estimates result in out-of-distribution detection that outperforms the capabilities of previous Bayesian networks and reconstruction-based approaches that learn normative distributions. To facilitate future work our code is publicly available.

Continual Learning of Dynamical Systems with Competitive Federated Reservoir Computing

Machine learning recently proved efficient in learning differential equations and dynamical systems from data. However, the data is commonly assumed to originate from a single never-changing system. In contrast, when modeling real-world dynamical processes, the data distribution often shifts due to changes in the underlying system dynamics. Continual learning of these processes aims to rapidly adapt to abrupt system changes without forgetting previous dynamical regimes. This work proposes an approach to continual learning based on reservoir computing, a state-of-the-art method for training recurrent neural networks on complex spatiotemporal dynamical systems. Reservoir computing fixes the recurrent network weights - hence these cannot be forgotten - and only updates linear projection heads to the output. We propose to train multiple competitive prediction heads concurrently. Inspired by neuroscience's predictive coding, only the most predictive heads activate, laterally inhibiting and thus protecting the inactive heads from forgetting induced by interfering parameter updates. We show that this multi-head reservoir minimizes interference and catastrophic forgetting on several dynamical systems, including the Van-der-Pol oscillator, the chaotic Lorenz attractor, and the high-dimensional Lorenz-96 weather model. Our results suggest that reservoir computing is a promising candidate framework for the continual learning of dynamical systems. We provide our code for data generation, method, and comparisons at \url{https://github.com/leonardbereska/multiheadreservoir}.

Accurate and fast identification of minimally prepared bacteria phenotypes using Raman spectroscopy assisted by machine learning

The worldwide increase of antimicrobial resistance (AMR) is a serious threat to human health. To avert the spread of AMR, fast reliable diagnostics tools that facilitate optimal antibiotic stewardship are an unmet need. In this regard, Raman spectroscopy promises rapid label- and culture-free identification and antimicrobial susceptibility testing (AST) in a single step. However, even though many Raman-based bacteria-identification and AST studies have demonstrated impressive results, some shortcomings must be addressed. To bridge the gap between proof-of-concept studies and clinical application, we have developed machine learning techniques in combination with a novel data-augmentation algorithm, for fast identification of minimally prepared bacteria phenotypes and the distinctions of methicillin-resistant (MR) from methicillin-susceptible (MS) bacteria. For this we have implemented a spectral transformer model for hyper-spectral Raman images of bacteria. We show that our model outperforms the standard convolutional neural network models on a multitude of classification problems, both in terms of accuracy and in terms of training time. We attain more than 96$\%$ classification accuracy on a dataset consisting of 15 different classes and 95.6$\%$ classification accuracy for six MR-MS bacteria species. More importantly, our results are obtained using only fast and easy-to-produce training and test data

A penalisation method for batch multi-objective Bayesian optimisation with application in heat exchanger design

We present HIghly Parallelisable Pareto Optimisation (HIPPO) -- a batch acquisition function that enables multi-objective Bayesian optimisation methods to efficiently exploit parallel processing resources. Multi-Objective Bayesian Optimisation (MOBO) is a very efficient tool for tackling expensive black-box problems. However, most MOBO algorithms are designed as purely sequential strategies, and existing batch approaches are prohibitively expensive for all but the smallest of batch sizes. We show that by encouraging batch diversity through penalising evaluations with similar predicted objective values, HIPPO is able to cheaply build large batches of informative points. Our extensive experimental validation demonstrates that HIPPO is at least as efficient as existing alternatives whilst incurring an order of magnitude lower computational overhead and scaling easily to batch sizes considerably higher than currently supported in the literature. Additionally, we demonstrate the application of HIPPO to a challenging heat exchanger design problem, stressing the real-world utility of our highly parallelisable approach to MOBO.

Path Integral Stochastic Optimal Control for Sampling Transition Paths

We consider the problem of Sampling Transition Paths. Given two metastable conformational states of a molecular system, eg. a folded and unfolded protein, we aim to sample the most likely transition path between the two states. Sampling such a transition path is computationally expensive due to the existence of high free energy barriers between the two states. To circumvent this, previous work has focused on simplifying the trajectories to occur along specific molecular descriptors called Collective Variables (CVs). However, finding CVs is not trivial and requires chemical intuition. For larger molecules, where intuition is not sufficient, using these CV-based methods biases the transition along possibly irrelevant dimensions. Instead, this work proposes a method for sampling transition paths that consider the entire geometry of the molecules. To achieve this, we first relate the problem to recent work on the Schrodinger bridge problem and stochastic optimal control. Using this relation, we construct a method that takes into account important characteristics of molecular systems such as second-order dynamics and invariance to rotations and translations. We demonstrate our method on the commonly studied Alanine Dipeptide, but also consider larger proteins such as Polyproline and Chignolin.

Humans are not Boltzmann Distributions: Challenges and Opportunities for Modelling Human Feedback and Interaction in Reinforcement Learning

Reinforcement learning (RL) commonly assumes access to well-specified reward functions, which many practical applications do not provide. Instead, recently, more work has explored learning what to do from interacting with humans. So far, most of these approaches model humans as being (nosily) rational and, in particular, giving unbiased feedback. We argue that these models are too simplistic and that RL researchers need to develop more realistic human models to design and evaluate their algorithms. In particular, we argue that human models have to be personal, contextual, and dynamic. This paper calls for research from different disciplines to address key questions about how humans provide feedback to AIs and how we can build more robust human-in-the-loop RL systems.

Insights into Deep Non-linear Filters for Improved Multi-channel Speech Enhancement

The key advantage of using multiple microphones for speech enhancement is that spatial filtering can be used to complement the tempo-spectral processing. In a traditional setting, linear spatial filtering (beamforming) and single-channel post-filtering are commonly performed separately. In contrast, there is a trend towards employing deep neural networks (DNNs) to learn a joint spatial and tempo-spectral non-linear filter, which means that the restriction of a linear processing model and that of a separate processing of spatial and tempo-spectral information can potentially be overcome. However, the internal mechanisms that lead to good performance of such data-driven filters for multi-channel speech enhancement are not well understood. Therefore, in this work, we analyse the properties of a non-linear spatial filter realized by a DNN as well as its interdependency with temporal and spectral processing by carefully controlling the information sources (spatial, spectral, and temporal) available to the network. We confirm the superiority of a non-linear spatial processing model, which outperforms an oracle linear spatial filter in a challenging speaker extraction scenario for a low number of microphones by 0.24 POLQA score. Our analyses reveal that in particular spectral information should be processed jointly with spatial information as this increases the spatial selectivity of the filter. Our systematic evaluation then leads to a simple network architecture, that outperforms state-of-the-art network architectures on a speaker extraction task by 0.22 POLQA score and by 0.32 POLQA score on the CHiME3 data.

Supernova Light Curves Approximation based on Neural Network Models

Photometric data-driven classification of supernovae becomes a challenge due to the appearance of real-time processing of big data in astronomy. Recent studies have demonstrated the superior quality of solutions based on various machine learning models. These models learn to classify supernova types using their light curves as inputs. Preprocessing these curves is a crucial step that significantly affects the final quality. In this talk, we study the application of multilayer perceptron (MLP), bayesian neural network (BNN), and normalizing flows (NF) to approximate observations for a single light curve. We use these approximations as inputs for supernovae classification models and demonstrate that the proposed methods outperform the state-of-the-art based on Gaussian processes applying to the Zwicky Transient Facility Bright Transient Survey light curves. MLP demonstrates similar quality as Gaussian processes and speed increase. Normalizing Flows exceeds Gaussian processes in terms of approximation quality as well.

PARTICUL: Part Identification with Confidence measure using Unsupervised Learning

In this paper, we present PARTICUL, a novel algorithm for unsupervised learning of part detectors from datasets used in fine-grained recognition. It exploits the macro-similarities of all images in the training set in order to mine for recurring patterns in the feature space of a pre-trained convolutional neural network. We propose new objective functions enforcing the locality and unicity of the detected parts. Additionally, we embed our detectors with a confidence measure based on correlation scores, allowing the system to estimate the visibility of each part. We apply our method on two public fine-grained datasets (Caltech-UCSD Bird 200 and Stanford Cars) and show that our detectors can consistently highlight parts of the object while providing a good measure of the confidence in their prediction. We also demonstrate that these detectors can be directly used to build part-based fine-grained classifiers that provide a good compromise between the transparency of prototype-based approaches and the performance of non-interpretable methods.

A Multilingual Dataset of COVID-19 Vaccination Attitudes on Twitter

Vaccine hesitancy is considered as one main cause of the stagnant uptake ratio of COVID-19 vaccines in Europe and the US where vaccines are sufficiently supplied. Fast and accurate grasp of public attitudes toward vaccination is critical to address vaccine hesitancy, and social media platforms have proved to be an effective source of public opinions. In this paper, we describe the collection and release of a dataset of tweets related to COVID-19 vaccines. This dataset consists of the IDs of 2,198,090 tweets collected from Western Europe, 17,934 of which are annotated with the originators' vaccination stances. Our annotation will facilitate using and developing data-driven models to extract vaccination attitudes from social media posts and thus further confirm the power of social media in public health surveillance. To lay the groundwork for future research, we not only perform statistical analysis and visualisation of our dataset, but also evaluate and compare the performance of established text-based benchmarks in vaccination stance extraction. We demonstrate one potential use of our data in practice in tracking the temporal changes of public COVID-19 vaccination attitudes.

Consistency-preserving Visual Question Answering in Medical Imaging

Visual Question Answering (VQA) models take an image and a natural-language question as input and infer the answer to the question. Recently, VQA systems in medical imaging have gained popularity thanks to potential advantages such as patient engagement and second opinions for clinicians. While most research efforts have been focused on improving architectures and overcoming data-related limitations, answer consistency has been overlooked even though it plays a critical role in establishing trustworthy models. In this work, we propose a novel loss function and corresponding training procedure that allows the inclusion of relations between questions into the training process. Specifically, we consider the case where implications between perception and reasoning questions are known a-priori. To show the benefits of our approach, we evaluate it on the clinically relevant task of Diabetic Macular Edema (DME) staging from fundus imaging. Our experiments show that our method outperforms state-of-the-art baselines, not only by improving model consistency, but also in terms of overall model accuracy. Our code and data are available at https://github.com/sergiotasconmorales/consistency_vqa.

Diffusion Deformable Model for 4D Temporal Medical Image Generation

Temporal volume images with 3D+t (4D) information are often used in medical imaging to statistically analyze temporal dynamics or capture disease progression. Although deep-learning-based generative models for natural images have been extensively studied, approaches for temporal medical image generation such as 4D cardiac volume data are limited. In this work, we present a novel deep learning model that generates intermediate temporal volumes between source and target volumes. Specifically, we propose a diffusion deformable model (DDM) by adapting the denoising diffusion probabilistic model that has recently been widely investigated for realistic image generation. Our proposed DDM is composed of the diffusion and the deformation modules so that DDM can learn spatial deformation information between the source and target volumes and provide a latent code for generating intermediate frames along a geodesic path. Once our model is trained, the latent code estimated from the diffusion module is simply interpolated and fed into the deformation module, which enables DDM to generate temporal frames along the continuous trajectory while preserving the topology of the source image. We demonstrate the proposed method with the 4D cardiac MR image generation between the diastolic and systolic phases for each subject. Compared to the existing deformation methods, our DDM achieves high performance on temporal volume generation.