Datasets:

CShorten

/

Last-Week-on-ML-ArXiv

like 0

Theoretical analysis of Adam using hyperparameters close to one without Lipschitz smoothness

Convergence and convergence rate analyses of adaptive methods, such as Adaptive Moment Estimation (Adam) and its variants, have been widely studied for nonconvex optimization. The analyses are based on assumptions that the expected or empirical average loss function is Lipschitz smooth (i.e., its gradient is Lipschitz continuous) and the learning rates depend on the Lipschitz constant of the Lipschitz continuous gradient. Meanwhile, numerical evaluations of Adam and its variants have clarified that using small constant learning rates without depending on the Lipschitz constant and hyperparameters ($\beta_1$ and $\beta_2$) close to one is advantageous for training deep neural networks. Since computing the Lipschitz constant is NP-hard, the Lipschitz smoothness condition would be unrealistic. This paper provides theoretical analyses of Adam without assuming the Lipschitz smoothness condition in order to bridge the gap between theory and practice. The main contribution is to show theoretical evidence that Adam using small learning rates and hyperparameters close to one performs well, whereas the previous theoretical results were all for hyperparameters close to zero. Our analysis also leads to the finding that Adam performs well with large batch sizes. Moreover, we show that Adam performs well when it uses diminishing learning rates and hyperparameters close to one.

Sum-of-Squares Relaxations for Information Theory and Variational Inference

We consider extensions of the Shannon relative entropy, referred to as f-divergences. Three classical related computational problems are typically associated with these divergences: (a) estimation from moments, (b) computing normalizing integrals, and (c) variational inference in probabilistic models. These problems are related to one another through convex duality, and for all them, there are many applications throughout data science, and we aim for computationally tractable approximation algorithms that preserve properties of the original problem such as potential convexity or monotonicity. In order to achieve this, we derive a sequence of convex relaxations for computing these divergences from non-centered covariance matrices associated with a given feature vector: starting from the typically non-tractable optimal lower-bound, we consider an additional relaxation based on ''sums-of-squares'', which is is now computable in polynomial time as a semidefinite program, as well as further computationally more efficient relaxations based on spectral information divergences from quantum information theory. For all of the tasks above, beyond proposing new relaxations, we derive tractable algorithms based on augmented Lagrangians and first-order methods, and we present illustrations on multivariate trigonometric polynomials and functions on the Boolean hypercube.

Deep-Learning vs Regression: Prediction of Tourism Flow with Limited Data

Modern tourism in the 21st century is facing numerous challenges. One of these challenges is the rapidly growing number of tourists in space limited regions such as historical city centers, museums or geographical bottlenecks like narrow valleys. In this context, a proper and accurate prediction of tourism volume and tourism flow within a certain area is important and critical for visitor management tasks such as visitor flow control and prevention of overcrowding. Static flow control methods like limiting access to hotspots or using conventional low level controllers could not solve the problem yet. In this paper, we empirically evaluate the performance of several state-of-the-art deep-learning methods in the field of visitor flow prediction with limited data by using available granular data supplied by a tourism region and comparing the results to ARIMA, a classical statistical method. Our results show that deep-learning models yield better predictions compared to the ARIMA method, while both featuring faster inference times and being able to incorporate additional input features.

Wideband Audio Waveform Evaluation Networks: Efficient, Accurate Estimation of Speech Qualities

Wideband Audio Waveform Evaluation Networks (WAWEnets) are convolutional neural networks that operate directly on wideband audio waveforms in order to produce evaluations of those waveforms. In the present work these evaluations give qualities of telecommunications speech (e.g., noisiness, intelligibility, overall speech quality). WAWEnets are no-reference networks because they do not require ``reference'' (original or undistorted) versions of the waveforms they evaluate. Our initial WAWEnet publication introduced four WAWEnets and each emulated the output of an established full-reference speech quality or intelligibility estimation algorithm. We have updated the WAWEnet architecture to be more efficient and effective. Here we present a single WAWEnet that closely tracks seven different quality and intelligibility values. We create a second network that additionally tracks four subjective speech quality dimensions. We offer a third network that focuses on just subjective quality scores and achieves very high levels of agreement. This work has leveraged 334 hours of speech in 13 languages, over two million full-reference target values and over 93,000 subjective mean opinion scores. We also interpret the operation of WAWEnets and identify the key to their operation using the language of signal processing: ReLUs strategically move spectral information from non-DC components into the DC component. The DC values of 96 output signals define a vector in a 96-D latent space and this vector is then mapped to a quality or intelligibility value for the input waveform.

Multifamily Malware Models

When training a machine learning model, there is likely to be a tradeoff between accuracy and the diversity of the dataset. Previous research has shown that if we train a model to detect one specific malware family, we generally obtain stronger results as compared to a case where we train a single model on multiple diverse families. However, during the detection phase, it would be more efficient to have a single model that can reliably detect multiple families, rather than having to score each sample against multiple models. In this research, we conduct experiments based on byte $n$-gram features to quantify the relationship between the generality of the training dataset and the accuracy of the corresponding machine learning models, all within the context of the malware detection problem. We find that neighborhood-based algorithms generalize surprisingly well, far outperforming the other machine learning techniques considered.

The Performance of Wasserstein Distributionally Robust M-Estimators in High Dimensions

Wasserstein distributionally robust optimization has recently emerged as a powerful framework for robust estimation, enjoying good out-of-sample performance guarantees, well-understood regularization effects, and computationally tractable dual reformulations. In such framework, the estimator is obtained by minimizing the worst-case expected loss over all probability distributions which are close, in a Wasserstein sense, to the empirical distribution. In this paper, we propose a Wasserstein distributionally robust M-estimation framework to estimate an unknown parameter from noisy linear measurements, and we focus on the important and challenging task of analyzing the squared error performance of such estimators. Our study is carried out in the modern high-dimensional proportional regime, where both the ambient dimension and the number of samples go to infinity, at a proportional rate which encodes the under/over-parametrization of the problem. Under an isotropic Gaussian features assumption, we show that the squared error can be recover as the solution of a convex-concave optimization problem which, surprinsingly, involves at most four scalar variables. To the best of our knowledge, this is the first work to study this problem in the context of Wasserstein distributionally robust M-estimation.

Finite Littlestone Dimension Implies Finite Information Complexity

We prove that every online learnable class of functions of Littlestone dimension $d$ admits a learning algorithm with finite information complexity. Towards this end, we use the notion of a globally stable algorithm. Generally, the information complexity of such a globally stable algorithm is large yet finite, roughly exponential in $d$. We also show there is room for improvement; for a canonical online learnable class, indicator functions of affine subspaces of dimension $d$, the information complexity can be upper bounded logarithmically in $d$.

Knowledge-aware Neural Collective Matrix Factorization for Cross-domain Recommendation

Cross-domain recommendation (CDR) can help customers find more satisfying items in different domains. Existing CDR models mainly use common users or mapping functions as bridges between domains but have very limited exploration in fully utilizing extra knowledge across domains. In this paper, we propose to incorporate the knowledge graph (KG) for CDR, which enables items in different domains to share knowledge. To this end, we first construct a new dataset AmazonKG4CDR from the Freebase KG and a subset (two domain pairs: movies-music, movie-book) of Amazon Review Data. This new dataset facilitates linking knowledge to bridge within- and cross-domain items for CDR. Then we propose a new framework, KG-aware Neural Collective Matrix Factorization (KG-NeuCMF), leveraging KG to enrich item representations. It first learns item embeddings by graph convolutional autoencoder to capture both domain-specific and domain-general knowledge from adjacent and higher-order neighbours in the KG. Then, we maximize the mutual information between item embeddings learned from the KG and user-item matrix to establish cross-domain relationships for better CDR. Finally, we conduct extensive experiments on the newly constructed dataset and demonstrate that our model significantly outperforms the best-performing baselines.

Sample compression schemes for balls in graphs

One of the open problems in machine learning is whether any set-family of VC-dimension $d$ admits a sample compression scheme of size~$O(d)$. In this paper, we study this problem for balls in graphs. For balls of arbitrary radius $r$, we design proper sample compression schemes of size $2$ for trees, of size $3$ for cycles, of size $4$ for interval graphs, of size $6$ for trees of cycles, and of size $22$ for cube-free median graphs. For balls of a given radius, we design proper labeled sample compression schemes of size $2$ for trees and of size $4$ for interval graphs. We also design approximate sample compression schemes of size 2 for balls of $\delta$-hyperbolic graphs.

Attack Agnostic Dataset: Towards Generalization and Stabilization of Audio DeepFake Detection

Audio DeepFakes allow the creation of high-quality, convincing utterances and therefore pose a threat due to its potential applications such as impersonation or fake news. Methods for detecting these manipulations should be characterized by good generalization and stability leading to robustness against attacks conducted with techniques that are not explicitly included in the training. In this work, we introduce Attack Agnostic Dataset - a combination of two audio DeepFakes and one anti-spoofing datasets that, thanks to the disjoint use of attacks, can lead to better generalization of detection methods. We present a thorough analysis of current DeepFake detection methods and consider different audio features (front-ends). In addition, we propose a model based on LCNN with LFCC and mel-spectrogram front-end, which not only is characterized by a good generalization and stability results but also shows improvement over LFCC-based mode - we decrease standard deviation on all folds and EER in two folds by up to 5%.

Cracking nuts with a sledgehammer: when modern graph neural networks do worse than classical greedy algorithms

The recent work ``Combinatorial Optimization with Physics-Inspired Graph Neural Networks'' [Nat Mach Intell 4 (2022) 367] introduces a physics-inspired unsupervised Graph Neural Network (GNN) to solve combinatorial optimization problems on sparse graphs. To test the performances of these GNNs, the authors of the work show numerical results for two fundamental problems: maximum cut and maximum independent set (MIS). They conclude that "the graph neural network optimizer performs on par or outperforms existing solvers, with the ability to scale beyond the state of the art to problems with millions of variables." In this comment, we show that a simple greedy algorithm, running in almost linear time, can find solutions for the MIS problem of much better quality than the GNN. The greedy algorithm is faster by a factor of $10^4$ with respect to the GNN for problems with a million variables. We do not see any good reason for solving the MIS with these GNN, as well as for using a sledgehammer to crack nuts. In general, many claims of superiority of neural networks in solving combinatorial problems are at risk of being not solid enough, since we lack standard benchmarks based on really hard problems. We propose one of such hard benchmarks, and we hope to see future neural network optimizers tested on these problems before any claim of superiority is made.

Human-AI Collaboration in Decision-Making: Beyond Learning to Defer

Human-AI collaboration (HAIC) in decision-making aims to create synergistic teaming between human decision-makers and AI systems. Learning to Defer (L2D) has been presented as a promising framework to determine who among humans and AI should take which decisions in order to optimize the performance and fairness of the combined system. Nevertheless, L2D entails several often unfeasible requirements, such as the availability of predictions from humans for every instance or ground-truth labels independent from said decision-makers. Furthermore, neither L2D nor alternative approaches tackle fundamental issues of deploying HAIC in real-world settings, such as capacity management or dealing with dynamic environments. In this paper, we aim to identify and review these and other limitations, pointing to where opportunities for future research in HAIC may lie.

Few-Shot Cross-Lingual TTS Using Transferable Phoneme Embedding

This paper studies a transferable phoneme embedding framework that aims to deal with the cross-lingual text-to-speech (TTS) problem under the few-shot setting. Transfer learning is a common approach when it comes to few-shot learning since training from scratch on few-shot training data is bound to overfit. Still, we find that the naive transfer learning approach fails to adapt to unseen languages under extremely few-shot settings, where less than 8 minutes of data is provided. We deal with the problem by proposing a framework that consists of a phoneme-based TTS model and a codebook module to project phonemes from different languages into a learned latent space. Furthermore, by utilizing phoneme-level averaged self-supervised learned features, we effectively improve the quality of synthesized speeches. Experiments show that using 4 utterances, which is about 30 seconds of data, is enough to synthesize intelligible speech when adapting to an unseen language using our framework.

Differentially Private Federated Combinatorial Bandits with Constraints

There is a rapid increase in the cooperative learning paradigm in online learning settings, i.e., federated learning (FL). Unlike most FL settings, there are many situations where the agents are competitive. Each agent would like to learn from others, but the part of the information it shares for others to learn from could be sensitive; thus, it desires its privacy. This work investigates a group of agents working concurrently to solve similar combinatorial bandit problems while maintaining quality constraints. Can these agents collectively learn while keeping their sensitive information confidential by employing differential privacy? We observe that communicating can reduce the regret. However, differential privacy techniques for protecting sensitive information makes the data noisy and may deteriorate than help to improve regret. Hence, we note that it is essential to decide when to communicate and what shared data to learn to strike a functional balance between regret and privacy. For such a federated combinatorial MAB setting, we propose a Privacy-preserving Federated Combinatorial Bandit algorithm, P-FCB. We illustrate the efficacy of P-FCB through simulations. We further show that our algorithm provides an improvement in terms of regret while upholding quality threshold and meaningful privacy guarantees.

An Empirical Study of Personalized Federated Learning

Federated learning is a distributed machine learning approach in which a single server and multiple clients collaboratively build machine learning models without sharing datasets on clients. A challenging issue of federated learning is data heterogeneity (i.e., data distributions may differ across clients). To cope with this issue, numerous federated learning methods aim at personalized federated learning and build optimized models for clients. Whereas existing studies empirically evaluated their own methods, the experimental settings (e.g., comparison methods, datasets, and client setting) in these studies differ from each other, and it is unclear which personalized federate learning method achieves the best performance and how much progress can be made by using these methods instead of standard (i.e., non-personalized) federated learning. In this paper, we benchmark the performance of existing personalized federated learning through comprehensive experiments to evaluate the characteristics of each method. Our experimental study shows that (1) there are no champion methods, (2) large data heterogeneity often leads to high accurate predictions, and (3) standard federated learning methods (e.g. FedAvg) with fine-tuning often outperform personalized federated learning methods. We open our benchmark tool FedBench for researchers to conduct experimental studies with various experimental settings.

Prisoners of Their Own Devices: How Models Induce Data Bias in Performative Prediction

The unparalleled ability of machine learning algorithms to learn patterns from data also enables them to incorporate biases embedded within. A biased model can then make decisions that disproportionately harm certain groups in society. Much work has been devoted to measuring unfairness in static ML environments, but not in dynamic, performative prediction ones, in which most real-world use cases operate. In the latter, the predictive model itself plays a pivotal role in shaping the distribution of the data. However, little attention has been heeded to relating unfairness to these interactions. Thus, to further the understanding of unfairness in these settings, we propose a taxonomy to characterize bias in the data, and study cases where it is shaped by model behaviour. Using a real-world account opening fraud detection case study as an example, we study the dangers to both performance and fairness of two typical biases in performative prediction: distribution shifts, and the problem of selective labels.

Discrete Morse Sandwich: Fast Computation of Persistence Diagrams for Scalar Data -- An Algorithm and A Benchmark

This paper introduces an efficient algorithm for persistence diagram computation, given an input piecewise linear scalar field f defined on a d-dimensional simplicial complex K, with $d \leq 3$. Our method extends the seminal "PairCells" algorithm by introducing three main accelerations. First, we express this algorithm within the setting of discrete Morse theory, which considerably reduces the number of input simplices to consider. Second, we introduce a stratification approach to the problem, that we call "sandwiching". Specifically, minima-saddle persistence pairs ($D_0(f)$) and saddle-maximum persistence pairs ($D_{d-1}(f)$) are efficiently computed by respectively processing with a Union-Find the unstable sets of 1-saddles and the stable sets of (d-1)-saddles. This fast processing of the dimensions 0 and (d-1) further reduces, and drastically, the number of critical simplices to consider for the computation of $D_1(f)$, the intermediate layer of the sandwich. Third, we document several performance improvements via shared-memory parallelism. We provide an open-source implementation of our algorithm for reproducibility purposes. We also contribute a reproducible benchmark package, which exploits three-dimensional data from a public repository and compares our algorithm to a variety of publicly available implementations. Extensive experiments indicate that our algorithm improves by two orders of magnitude the time performance of the seminal "PairCells" algorithm it extends. Moreover, it also improves memory footprint and time performance over a selection of 14 competing approaches, with a substantial gain over the fastest available approaches, while producing a strictly identical output. We illustrate the utility of our contributions with an application to the fast and robust extraction of persistent 1-dimensional generators on surfaces, volume data and high-dimensional point clouds.

Learning to Control Local Search for Combinatorial Optimization

Combinatorial optimization problems are encountered in many practical contexts such as logistics and production, but exact solutions are particularly difficult to find and usually NP-hard for considerable problem sizes. To compute approximate solutions, a zoo of generic as well as problem-specific variants of local search is commonly used. However, which variant to apply to which particular problem is difficult to decide even for experts. In this paper we identify three independent algorithmic aspects of such local search algorithms and formalize their sequential selection over an optimization process as Markov Decision Process (MDP). We design a deep graph neural network as policy model for this MDP, yielding a learned controller for local search called NeuroLS. Ample experimental evidence shows that NeuroLS is able to outperform both, well-known general purpose local search controllers from Operations Research as well as latest machine learning-based approaches.

A Representation Learning Framework for Property Graphs

Representation learning on graphs, also called graph embedding, has demonstrated its significant impact on a series of machine learning applications such as classification, prediction and recommendation. However, existing work has largely ignored the rich information contained in the properties (or attributes) of both nodes and edges of graphs in modern applications, e.g., those represented by property graphs. To date, most existing graph embedding methods either focus on plain graphs with only the graph topology, or consider properties on nodes only. We propose PGE, a graph representation learning framework that incorporates both node and edge properties into the graph embedding procedure. PGE uses node clustering to assign biases to differentiate neighbors of a node and leverages multiple data-driven matrices to aggregate the property information of neighbors sampled based on a biased strategy. PGE adopts the popular inductive model for neighborhood aggregation. We provide detailed analyses on the efficacy of our method and validate the performance of PGE by showing how PGE achieves better embedding results than the state-of-the-art graph embedding methods on benchmark applications such as node classification and link prediction over real-world datasets.

Measuring and Improving the Use of Graph Information in Graph Neural Networks

Graph neural networks (GNNs) have been widely used for representation learning on graph data. However, there is limited understanding on how much performance GNNs actually gain from graph data. This paper introduces a context-surrounding GNN framework and proposes two smoothness metrics to measure the quantity and quality of information obtained from graph data. A new GNN model, called CS-GNN, is then designed to improve the use of graph information based on the smoothness values of a graph. CS-GNN is shown to achieve better performance than existing methods in different types of real graphs.

Local Evaluation of Time Series Anomaly Detection Algorithms

In recent years, specific evaluation metrics for time series anomaly detection algorithms have been developed to handle the limitations of the classical precision and recall. However, such metrics are heuristically built as an aggregate of multiple desirable aspects, introduce parameters and wipe out the interpretability of the output. In this article, we first highlight the limitations of the classical precision/recall, as well as the main issues of the recent event-based metrics -- for instance, we show that an adversary algorithm can reach high precision and recall on almost any dataset under weak assumption. To cope with the above problems, we propose a theoretically grounded, robust, parameter-free and interpretable extension to precision/recall metrics, based on the concept of ``affiliation'' between the ground truth and the prediction sets. Our metrics leverage measures of duration between ground truth and predictions, and have thus an intuitive interpretation. By further comparison against random sampling, we obtain a normalized precision/recall, quantifying how much a given set of results is better than a random baseline prediction. By construction, our approach keeps the evaluation local regarding ground truth events, enabling fine-grained visualization and interpretation of algorithmic results. We compare our proposal against various public time series anomaly detection datasets, algorithms and metrics. We further derive theoretical properties of the affiliation metrics that give explicit expectations about their behavior and ensure robustness against adversary strategies.

Evaluating resampling methods on a real-life highly imbalanced online credit card payments dataset

Various problems of any credit card fraud detection based on machine learning come from the imbalanced aspect of transaction datasets. Indeed, the number of frauds compared to the number of regular transactions is tiny and has been shown to damage learning performances, e.g., at worst, the algorithm can learn to classify all the transactions as regular. Resampling methods and cost-sensitive approaches are known to be good candidates to leverage this issue of imbalanced datasets. This paper evaluates numerous state-of-the-art resampling methods on a large real-life online credit card payments dataset. We show they are inefficient because methods are intractable or because metrics do not exhibit substantial improvements. Our work contributes to this domain in (1) that we compare many state-of-the-art resampling methods on a large-scale dataset and in (2) that we use a real-life online credit card payments dataset.

Compressing Features for Learning with Noisy Labels

Supervised learning can be viewed as distilling relevant information from input data into feature representations. This process becomes difficult when supervision is noisy as the distilled information might not be relevant. In fact, recent research shows that networks can easily overfit all labels including those that are corrupted, and hence can hardly generalize to clean datasets. In this paper, we focus on the problem of learning with noisy labels and introduce compression inductive bias to network architectures to alleviate this over-fitting problem. More precisely, we revisit one classical regularization named Dropout and its variant Nested Dropout. Dropout can serve as a compression constraint for its feature dropping mechanism, while Nested Dropout further learns ordered feature representations w.r.t. feature importance. Moreover, the trained models with compression regularization are further combined with Co-teaching for performance boost. Theoretically, we conduct bias-variance decomposition of the objective function under compression regularization. We analyze it for both single model and Co-teaching. This decomposition provides three insights: (i) it shows that over-fitting is indeed an issue for learning with noisy labels; (ii) through an information bottleneck formulation, it explains why the proposed feature compression helps in combating label noise; (iii) it gives explanations on the performance boost brought by incorporating compression regularization into Co-teaching. Experiments show that our simple approach can have comparable or even better performance than the state-of-the-art methods on benchmarks with real-world label noise including Clothing1M and ANIMAL-10N. Our implementation is available at https://yingyichen-cyy.github.io/CompressFeatNoisyLabels/.

Extracting Weighted Finite Automata from Recurrent Neural Networks for Natural Languages

Recurrent Neural Networks (RNNs) have achieved tremendous success in sequential data processing. However, it is quite challenging to interpret and verify RNNs' behaviors directly. To this end, many efforts have been made to extract finite automata from RNNs. Existing approaches such as exact learning are effective in extracting finite-state models to characterize the state dynamics of RNNs for formal languages, but are limited in the scalability to process natural languages. Compositional approaches that are scablable to natural languages fall short in extraction precision. In this paper, we identify the transition sparsity problem that heavily impacts the extraction precision. To address this problem, we propose a transition rule extraction approach, which is scalable to natural language processing models and effective in improving extraction precision. Specifically, we propose an empirical method to complement the missing rules in the transition diagram. In addition, we further adjust the transition matrices to enhance the context-aware ability of the extracted weighted finite automaton (WFA). Finally, we propose two data augmentation tactics to track more dynamic behaviors of the target RNN. Experiments on two popular natural language datasets show that our method can extract WFA from RNN for natural language processing with better precision than existing approaches.

Revisiting Architecture-aware Knowledge Distillation: Smaller Models and Faster Search

Knowledge Distillation (KD) has recently emerged as a popular method for compressing neural networks. In recent studies, generalized distillation methods that find parameters and architectures of student models at the same time have been proposed. Still, this search method requires a lot of computation to search for architectures and has the disadvantage of considering only convolutional blocks in their search space. This paper introduces a new algorithm, coined as Trust Region Aware architecture search to Distill knowledge Effectively (TRADE), that rapidly finds effective student architectures from several state-of-the-art architectures using trust region Bayesian optimization approach. Experimental results show our proposed TRADE algorithm consistently outperforms both the conventional NAS approach and pre-defined architecture under KD training.

Transfer learning for ensembles: reducing computation time and keeping the diversity

Transferring a deep neural network trained on one problem to another requires only a small amount of data and little additional computation time. The same behaviour holds for ensembles of deep learning models typically superior to a single model. However, a transfer of deep neural networks ensemble demands relatively high computational expenses. The probability of overfitting also increases. Our approach for the transfer learning of ensembles consists of two steps: (a) shifting weights of encoders of all models in the ensemble by a single shift vector and (b) doing a tiny fine-tuning for each individual model afterwards. This strategy leads to a speed-up of the training process and gives an opportunity to add models to an ensemble with significantly reduced training time using the shift vector. We compare different strategies by computation time, the accuracy of an ensemble, uncertainty estimation and disagreement and conclude that our approach gives competitive results using the same computation complexity in comparison with the traditional approach. Also, our method keeps the ensemble's models' diversity higher.

Dynamic-Group-Aware Networks for Multi-Agent Trajectory Prediction with Relational Reasoning

Demystifying the interactions among multiple agents from their past trajectories is fundamental to precise and interpretable trajectory prediction. However, previous works mainly consider static, pair-wise interactions with limited relational reasoning. To promote more comprehensive interaction modeling and relational reasoning, we propose DynGroupNet, a dynamic-group-aware network, which can i) model time-varying interactions in highly dynamic scenes; ii) capture both pair-wise and group-wise interactions; and iii) reason both interaction strength and category without direct supervision. Based on DynGroupNet, we further design a prediction system to forecast socially plausible trajectories with dynamic relational reasoning. The proposed prediction system leverages the Gaussian mixture model, multiple sampling and prediction refinement to promote prediction diversity, training stability and trajectory smoothness, respectively. Extensive experiments show that: 1)DynGroupNet can capture time-varying group behaviors, infer time-varying interaction category and interaction strength during trajectory prediction without any relation supervision on physical simulation datasets; 2)DynGroupNet outperforms the state-of-the-art trajectory prediction methods by a significant improvement of 22.6%/28.0%, 26.9%/34.9%, 5.1%/13.0% in ADE/FDE on the NBA, NFL Football and SDD datasets and achieve the state-of-the-art performance on the ETH-UCY dataset.

Enhancing Stochastic Petri Net-based Remaining Time Prediction using k-Nearest Neighbors

Reliable remaining time prediction of ongoing business processes is a highly relevant topic. One example is order delivery, a key competitive factor in e.g. retailing as it is a main driver of customer satisfaction. For realising timely delivery, an accurate prediction of the remaining time of the delivery process is crucial. Within the field of process mining, a wide variety of remaining time prediction techniques have already been proposed. In this work, we extend remaining time prediction based on stochastic Petri nets with generally distributed transitions with k-nearest neighbors. The k-nearest neighbors algorithm is performed on simple vectors storing the time passed to complete previous activities. By only taking a subset of instances, a more representative and stable stochastic Petri Net is obtained, leading to more accurate time predictions. We discuss the technique and its basic implementation in Python and use different real world data sets to evaluate the predictive power of our extension. These experiments show clear advantages in combining both techniques with regard to predictive power.

AdaSparse: Learning Adaptively Sparse Structures for Multi-Domain Click-Through Rate Prediction

Click-through rate (CTR) prediction is a fundamental technique in recommendation and advertising systems. Recent studies have proved that learning a unified model to serve multiple domains is effective to improve the overall performance. However, it is still challenging to improve generalization across domains under limited training data, and hard to deploy current solutions due to their computational complexity. In this paper, we propose a simple yet effective framework AdaSparse for multi-domain CTR prediction, which learns adaptively sparse structure for each domain, achieving better generalization across domains with lower computational cost. In AdaSparse, we introduce domain-aware neuron-level weighting factors to measure the importance of neurons, with that for each domain our model can prune redundant neurons to improve generalization. We further add flexible sparsity regularizations to control the sparsity ratio of learned structures. Offline and online experiments show that AdaSparse outperforms previous multi-domain CTR models significantly.

A mixed formulation for physics-informed neural networks as a potential solver for engineering problems in heterogeneous domains: comparison with finite element method

Physics-informed neural networks (PINNs) are capable of finding the solution for a given boundary value problem. We employ several ideas from the finite element method (FEM) to enhance the performance of existing PINNs in engineering problems. The main contribution of the current work is to promote using the spatial gradient of the primary variable as an output from separated neural networks. Later on, the strong form which has a higher order of derivatives is applied to the spatial gradients of the primary variable as the physical constraint. In addition, the so-called energy form of the problem is applied to the primary variable as an additional constraint for training. The proposed approach only required up to first-order derivatives to construct the physical loss functions. We discuss why this point is beneficial through various comparisons between different models. The mixed formulation-based PINNs and FE methods share some similarities. While the former minimizes the PDE and its energy form at given collocation points utilizing a complex nonlinear interpolation through a neural network, the latter does the same at element nodes with the help of shape functions. We focus on heterogeneous solids to show the capability of deep learning for predicting the solution in a complex environment under different boundary conditions. The performance of the proposed PINN model is checked against the solution from FEM on two prototype problems: elasticity and the Poisson equation (steady-state diffusion problem). We concluded that by properly designing the network architecture in PINN, the deep learning model has the potential to solve the unknowns in a heterogeneous domain without any available initial data from other sources. Finally, discussions are provided on the combination of PINN and FEM for a fast and accurate design of composite materials in future developments.

Modeling Content Creator Incentives on Algorithm-Curated Platforms

Content creators compete for user attention. Their reach crucially depends on algorithmic choices made by developers on online platforms. To maximize exposure, many creators adapt strategically, as evidenced by examples like the sprawling search engine optimization industry. This begets competition for the finite user attention pool. We formalize these dynamics in what we call an exposure game, a model of incentives induced by algorithms including modern factorization and (deep) two-tower architectures. We prove that seemingly innocuous algorithmic choices -- e.g., non-negative vs. unconstrained factorization -- significantly affect the existence and character of (Nash) equilibria in exposure games. We proffer use of creator behavior models like ours for an (ex-ante) pre-deployment audit. Such an audit can identify misalignment between desirable and incentivized content, and thus complement post-hoc measures like content filtering and moderation. To this end, we propose tools for numerically finding equilibria in exposure games, and illustrate results of an audit on the MovieLens and LastFM datasets. Among else, we find that the strategically produced content exhibits strong dependence between algorithmic exploration and content diversity, and between model expressivity and bias towards gender-based user and creator groups.

SpeechEQ: Speech Emotion Recognition based on Multi-scale Unified Datasets and Multitask Learning

Speech emotion recognition (SER) has many challenges, but one of the main challenges is that each framework does not have a unified standard. In this paper, we propose SpeechEQ, a framework for unifying SER tasks based on a multi-scale unified metric. This metric can be trained by Multitask Learning (MTL), which includes two emotion recognition tasks of Emotion States Category (EIS) and Emotion Intensity Scale (EIS), and two auxiliary tasks of phoneme recognition and gender recognition. For this framework, we build a Mandarin SER dataset - SpeechEQ Dataset (SEQD). We conducted experiments on the public CASIA and ESD datasets in Mandarin, which exhibit that our method outperforms baseline methods by a relatively large margin, yielding 8.0\% and 6.5\% improvement in accuracy respectively. Additional experiments on IEMOCAP with four emotion categories (i.e., angry, happy, sad, and neutral) also show the proposed method achieves a state-of-the-art of both weighted accuracy (WA) of 78.16% and unweighted accuracy (UA) of 77.47%.

Zero Stability Well Predicts Performance of Convolutional Neural Networks

The question of what kind of convolutional neural network (CNN) structure performs well is fascinating. In this work, we move toward the answer with one more step by connecting zero stability and model performance. Specifically, we found that if a discrete solver of an ordinary differential equation is zero stable, the CNN corresponding to that solver performs well. We first give the interpretation of zero stability in the context of deep learning and then investigate the performance of existing first- and second-order CNNs under different zero-stable circumstances. Based on the preliminary observation, we provide a higher-order discretization to construct CNNs and then propose a zero-stable network (ZeroSNet). To guarantee zero stability of the ZeroSNet, we first deduce a structure that meets consistency conditions and then give a zero stable region of a training-free parameter. By analyzing the roots of a characteristic equation, we theoretically obtain the optimal coefficients of feature maps. Empirically, we present our results from three aspects: We provide extensive empirical evidence of different depth on different datasets to show that the moduli of the characteristic equation's roots are the keys for the performance of CNNs that require historical features; Our experiments show that ZeroSNet outperforms existing CNNs which is based on high-order discretization; ZeroSNets show better robustness against noises on the input. The source code is available at \url{https://github.com/LongJin-lab/ZeroSNet}.

A Zero-Shot Classification Approach for a Word-Guessing Challenge

The Taboo Challenge competition, a task based on the well-known Taboo game, has been proposed to stimulate research in the AI field. The challenge requires building systems able to comprehend the implied inferences between the exchanged messages of guesser and describer agents. A describer sends pre-determined hints to guessers indirectly describing cities, and guessers are required to return the matching cities implied by the hints. Climbing up the scoring ledger requires the resolving of the highest amount of cities with the smallest amount of hints in a specified time frame. Here, we present TabooLM, a language-model approach that tackles the challenge based on a zero-shot setting. We start by presenting and comparing the results of this approach with three studies from the literature. The results show that our method achieves SOTA results on the Taboo challenge, suggesting that TabooLM can guess the implied cities faster and more accurately than existing approaches.

Learning Deep Input-Output Stable Dynamics

Learning stable dynamics from observed time-series data is an essential problem in robotics, physical modeling, and systems biology. Many of these dynamics are represented as an inputs-output system to communicate with the external environment. In this study, we focus on input-output stable systems, exhibiting robustness against unexpected stimuli and noise. We propose a method to learn nonlinear systems guaranteeing the input-output stability. Our proposed method utilizes the differentiable projection onto the space satisfying the Hamilton-Jacobi inequality to realize the input-output stability. The problem of finding this projection can be formulated as a quadratic constraint quadratic programming problem, and we derive the particular solution analytically. Also, we apply our method to a toy bistable model and the task of training a benchmark generated from a glucose-insulin simulator. The results show that the nonlinear system with neural networks by our method achieves the input-output stability, unlike naive neural networks. Our code is available at https://github.com/clinfo/DeepIOStability.

Split Localized Conformal Prediction

Conformal prediction is a simple and powerful tool that can quantify uncertainty without any distributional assumptions. However, existing methods can only provide an average coverage guarantee, which is not ideal compared to the stronger conditional coverage guarantee. Although achieving exact conditional coverage is proven to be impossible, approximating conditional coverage is still an important research direction. In this paper, we propose a modified non-conformity score by leveraging local approximation of the conditional distribution. The modified score inherits the spirit of split conformal methods, which is simple and efficient compared with full conformal methods but better approximates conditional coverage guarantee. Empirical results on various datasets, including a high dimension age regression on image, demonstrate that our method provides tighter intervals compared to existing methods.

Agreement-on-the-Line: Predicting the Performance of Neural Networks under Distribution Shift

Recently, Miller et al. showed that a model's in-distribution (ID) accuracy has a strong linear correlation with its out-of-distribution (OOD) accuracy on several OOD benchmarks -- a phenomenon they dubbed ''accuracy-on-the-line''. While a useful tool for model selection (i.e., the model most likely to perform the best OOD is the one with highest ID accuracy), this fact does not help estimate the actual OOD performance of models without access to a labeled OOD validation set. In this paper, we show a similar but surprising phenomenon also holds for the agreement between pairs of neural network classifiers: whenever accuracy-on-the-line holds, we observe that the OOD agreement between the predictions of any two pairs of neural networks (with potentially different architectures) also observes a strong linear correlation with their ID agreement. Furthermore, we observe that the slope and bias of OOD vs ID agreement closely matches that of OOD vs ID accuracy. This phenomenon, which we call agreement-on-the-line, has important practical applications: without any labeled data, we can predict the OOD accuracy of classifiers}, since OOD agreement can be estimated with just unlabeled data. Our prediction algorithm outperforms previous methods both in shifts where agreement-on-the-line holds and, surprisingly, when accuracy is not on the line. This phenomenon also provides new insights into deep neural networks: unlike accuracy-on-the-line, agreement-on-the-line appears to only hold for neural network classifiers.

RankSEG: A Consistent Ranking-based Framework for Segmentation

Segmentation has emerged as a fundamental field of computer vision and natural language processing, which assigns a label to every pixel/feature to extract regions of interest from an image/text. To evaluate the performance of segmentation, the Dice and IoU metrics are used to measure the degree of overlap between the ground truth and the predicted segmentation. In this paper, we establish a theoretical foundation of segmentation with respect to the Dice/IoU metrics, including the Bayes rule and Dice/IoU-calibration, analogous to classification-calibration or Fisher consistency in classification. We prove that the existing thresholding-based framework with most operating losses are not consistent with respect to the Dice/IoU metrics, and thus may lead to a suboptimal solution. To address this pitfall, we propose a novel consistent ranking-based framework, namely RankDice/RankIoU, inspired by plug-in rules of the Bayes segmentation rule. Three numerical algorithms with GPU parallel execution are developed to implement the proposed framework in large-scale and high-dimensional segmentation. We study statistical properties of the proposed framework. We show it is Dice-/IoU-calibrated, and its excess risk bounds and the rate of convergence are also provided. The numerical effectiveness of RankDice/mRankDice is demonstrated in various simulated examples and Fine-annotated CityScapes and Pascal VOC datasets with state-of-the-art deep learning architectures.

Adversarial Example Detection in Deployed Tree Ensembles

Tree ensembles are powerful models that are widely used. However, they are susceptible to adversarial examples, which are examples that purposely constructed to elicit a misprediction from the model. This can degrade performance and erode a user's trust in the model. Typically, approaches try to alleviate this problem by verifying how robust a learned ensemble is or robustifying the learning process. We take an alternative approach and attempt to detect adversarial examples in a post-deployment setting. We present a novel method for this task that works by analyzing an unseen example's output configuration, which is the set of predictions made by an ensemble's constituent trees. Our approach works with any additive tree ensemble and does not require training a separate model. We evaluate our approach on three different tree ensemble learners. We empirically show that our method is currently the best adversarial detection method for tree ensembles.

Improving Clinical Efficiency and Reducing Medical Errors through NLP-enabled diagnosis of Health Conditions from Transcription Reports

Misdiagnosis rates are one of the leading causes of medical errors in hospitals, affecting over 12 million adults across the US. To address the high rate of misdiagnosis, this study utilizes 4 NLP-based algorithms to determine the appropriate health condition based on an unstructured transcription report. From the Logistic Regression, Random Forest, LSTM, and CNNLSTM models, the CNN-LSTM model performed the best with an accuracy of 97.89%. We packaged this model into a authenticated web platform for accessible assistance to clinicians. Overall, by standardizing health care diagnosis and structuring transcription reports, our NLP platform drastically improves the clinical efficiency and accuracy of hospitals worldwide.

Leveraging Language for Accelerated Learning of Tool Manipulation

Robust and generalized tool manipulation requires an understanding of the properties and affordances of different tools. We investigate whether linguistic information about a tool (e.g., its geometry, common uses) can help control policies adapt faster to new tools for a given task. We obtain diverse descriptions of various tools in natural language and use pre-trained language models to generate their feature representations. We then perform language-conditioned meta-learning to learn policies that can efficiently adapt to new tools given their corresponding text descriptions. Our results demonstrate that combining linguistic information and meta-learning significantly accelerates tool learning in several manipulation tasks including pushing, lifting, sweeping, and hammering.

Uncertainty Calibration for Deep Audio Classifiers

Although deep Neural Networks (DNNs) have achieved tremendous success in audio classification tasks, their uncertainty calibration are still under-explored. A well-calibrated model should be accurate when it is certain about its prediction and indicate high uncertainty when it is likely to be inaccurate. In this work, we investigate the uncertainty calibration for deep audio classifiers. In particular, we empirically study the performance of popular calibration methods: (i) Monte Carlo Dropout, (ii) ensemble, (iii) focal loss, and (iv) spectral-normalized Gaussian process (SNGP), on audio classification datasets. To this end, we evaluate (i-iv) for the tasks of environment sound and music genre classification. Results indicate that uncalibrated deep audio classifiers may be over-confident, and SNGP performs the best and is very efficient on the two datasets of this paper.

On the Complexity of Adversarial Decision Making

A central problem in online learning and decision making -- from bandits to reinforcement learning -- is to understand what modeling assumptions lead to sample-efficient learning guarantees. We consider a general adversarial decision making framework that encompasses (structured) bandit problems with adversarial rewards and reinforcement learning problems with adversarial dynamics. Our main result is to show -- via new upper and lower bounds -- that the Decision-Estimation Coefficient, a complexity measure introduced by Foster et al. in the stochastic counterpart to our setting, is necessary and sufficient to obtain low regret for adversarial decision making. However, compared to the stochastic setting, one must apply the Decision-Estimation Coefficient to the convex hull of the class of models (or, hypotheses) under consideration. This establishes that the price of accommodating adversarial rewards or dynamics is governed by the behavior of the model class under convexification, and recovers a number of existing results -- both positive and negative. En route to obtaining these guarantees, we provide new structural results that connect the Decision-Estimation Coefficient to variants of other well-known complexity measures, including the Information Ratio of Russo and Van Roy and the Exploration-by-Optimization objective of Lattimore and Gy\"{o}rgy.

DPOAD: Differentially Private Outsourcing of Anomaly Detection through Iterative Sensitivity Learning

Outsourcing anomaly detection to third-parties can allow data owners to overcome resource constraints (e.g., in lightweight IoT devices), facilitate collaborative analysis (e.g., under distributed or multi-party scenarios), and benefit from lower costs and specialized expertise (e.g., of Managed Security Service Providers). Despite such benefits, a data owner may feel reluctant to outsource anomaly detection without sufficient privacy protection. To that end, most existing privacy solutions would face a novel challenge, i.e., preserving privacy usually requires the difference between data entries to be eliminated or reduced, whereas anomaly detection critically depends on that difference. Such a conflict is recently resolved under a local analysis setting with trusted analysts (where no outsourcing is involved) through moving the focus of differential privacy (DP) guarantee from "all" to only "benign" entries. In this paper, we observe that such an approach is not directly applicable to the outsourcing setting, because data owners do not know which entries are "benign" prior to outsourcing, and hence cannot selectively apply DP on data entries. Therefore, we propose a novel iterative solution for the data owner to gradually "disentangle" the anomalous entries from the benign ones such that the third-party analyst can produce accurate anomaly results with sufficient DP guarantee. We design and implement our Differentially Private Outsourcing of Anomaly Detection (DPOAD) framework, and demonstrate its benefits over baseline Laplace and PainFree mechanisms through experiments with real data from different application domains.

Automated Systems For Diagnosis of Dysgraphia in Children: A Survey and Novel Framework

Learning disabilities, which primarily interfere with the basic learning skills such as reading, writing and math, are known to affect around 10% of children in the world. The poor motor skills and motor coordination as part of the neurodevelopmental disorder can become a causative factor for the difficulty in learning to write (dysgraphia), hindering the academic track of an individual. The signs and symptoms of dysgraphia include but are not limited to irregular handwriting, improper handling of writing medium, slow or labored writing, unusual hand position, etc. The widely accepted assessment criterion for all the types of learning disabilities is the examination performed by medical experts. The few available artificial intelligence-powered screening systems for dysgraphia relies on the distinctive features of handwriting from the corresponding images.This work presents a review of the existing automated dysgraphia diagnosis systems for children in the literature. The main focus of the work is to review artificial intelligence-based systems for dysgraphia diagnosis in children. This work discusses the data collection method, important handwriting features, machine learning algorithms employed in the literature for the diagnosis of dysgraphia. Apart from that, this article discusses some of the non-artificial intelligence-based automated systems also. Furthermore, this article discusses the drawbacks of existing systems and proposes a novel framework for dysgraphia diagnosis.

Haul Road Mapping from GPS Traces

Automation in mining requires accurate maps of road networks on site. Because roads on open-cut mines are dynamic in nature and continuously changing, manually updating road maps is tedious and error-prone. This paper investigates the possibility of automatically deriving an accurate representation of the road network using GPS data available from haul trucks operating on site. We present an overview of approaches proposed in literature and test the performance of publicly available methods on GPS data collected from trucks operating on site. Based on shortcomings seen in all tested algorithms, a post-processing step is developed which geometrically analyses the created road map for artefacts typical of free-drive areas on mine sites and significantly improves the quality of the final road network graph.

A General Recipe for Likelihood-free Bayesian Optimization

The acquisition function, a critical component in Bayesian optimization (BO), can often be written as the expectation of a utility function under a surrogate model. However, to ensure that acquisition functions are tractable to optimize, restrictions must be placed on the surrogate model and utility function. To extend BO to a broader class of models and utilities, we propose likelihood-free BO (LFBO), an approach based on likelihood-free inference. LFBO directly models the acquisition function without having to separately perform inference with a probabilistic surrogate model. We show that computing the acquisition function in LFBO can be reduced to optimizing a weighted classification problem, where the weights correspond to the utility being chosen. By choosing the utility function for expected improvement (EI), LFBO outperforms various state-of-the-art black-box optimization methods on several real-world optimization problems. LFBO can also effectively leverage composite structures of the objective function, which further improves its regret by several orders of magnitude.

Monitoring Shortcut Learning using Mutual Information

The failure of deep neural networks to generalize to out-of-distribution data is a well-known problem and raises concerns about the deployment of trained networks in safety-critical domains such as healthcare, finance and autonomous vehicles. We study a particular kind of distribution shift $\unicode{x2013}$ shortcuts or spurious correlations in the training data. Shortcut learning is often only exposed when models are evaluated on real-world data that does not contain the same spurious correlations, posing a serious dilemma for AI practitioners to properly assess the effectiveness of a trained model for real-world applications. In this work, we propose to use the mutual information (MI) between the learned representation and the input as a metric to find where in training, the network latches onto shortcuts. Experiments demonstrate that MI can be used as a domain-agnostic metric for monitoring shortcut learning.

Normalized/Clipped SGD with Perturbation for Differentially Private Non-Convex Optimization

By ensuring differential privacy in the learning algorithms, one can rigorously mitigate the risk of large models memorizing sensitive training data. In this paper, we study two algorithms for this purpose, i.e., DP-SGD and DP-NSGD, which first clip or normalize \textit{per-sample} gradients to bound the sensitivity and then add noise to obfuscate the exact information. We analyze the convergence behavior of these two algorithms in the non-convex optimization setting with two common assumptions and achieve a rate $\mathcal{O}\left(\sqrt[4]{\frac{d\log(1/\delta)}{N^2\epsilon^2}}\right)$ of the gradient norm for a $d$-dimensional model, $N$ samples and $(\epsilon,\delta)$-DP, which improves over previous bounds under much weaker assumptions. Specifically, we introduce a regularizing factor in DP-NSGD and show that it is crucial in the convergence proof and subtly controls the bias and noise trade-off. Our proof deliberately handles the per-sample gradient clipping and normalization that are specified for the private setting. Empirically, we demonstrate that these two algorithms achieve similar best accuracy while DP-NSGD is comparatively easier to tune than DP-SGD and hence may help further save the privacy budget when accounting the tuning effort.

Towards Harnessing Feature Embedding for Robust Learning with Noisy Labels

The memorization effect of deep neural networks (DNNs) plays a pivotal role in recent label noise learning methods. To exploit this effect, the model prediction-based methods have been widely adopted, which aim to exploit the outputs of DNNs in the early stage of learning to correct noisy labels. However, we observe that the model will make mistakes during label prediction, resulting in unsatisfactory performance. By contrast, the produced features in the early stage of learning show better robustness. Inspired by this observation, in this paper, we propose a novel feature embedding-based method for deep learning with label noise, termed LabEl NoiseDilution (LEND). To be specific, we first compute a similarity matrix based on current embedded features to capture the local structure of training data. Then, the noisy supervision signals carried by mislabeled data are overwhelmed by nearby correctly labeled ones (\textit{i.e.}, label noise dilution), of which the effectiveness is guaranteed by the inherent robustness of feature embedding. Finally, the training data with diluted labels are further used to train a robust classifier. Empirically, we conduct extensive experiments on both synthetic and real-world noisy datasets by comparing our LEND with several representative robust learning approaches. The results verify the effectiveness of our LEND.

Long Range Language Modeling via Gated State Spaces

State space models have shown to be effective at modeling long range dependencies, specially on sequence classification tasks. In this work we focus on autoregressive sequence modeling over English books, Github source code and ArXiv mathematics articles. Based on recent developments around the effectiveness of gated activation functions, we propose a new layer named Gated State Space (GSS) and show that it trains significantly faster than the diagonal version of S4 (i.e. DSS) on TPUs, is fairly competitive with several well-tuned Transformer-based baselines and exhibits zero-shot generalization to longer inputs while being straightforward to implement. Finally, we show that leveraging self-attention to model local dependencies improves the performance of GSS even further.

Efficient Private SCO for Heavy-Tailed Data via Clipping

We consider stochastic convex optimization for heavy-tailed data with the guarantee of being differentially private (DP). Prior work on this problem is restricted to the gradient descent (GD) method, which is inefficient for large-scale problems. In this paper, we resolve this issue and derive the first high-probability bounds for private stochastic method with clipping. For general convex problems, we derive excess population risks $\Tilde{O}\left(\frac{d^{1/7}\sqrt{\ln\frac{(n \epsilon)^2}{\beta d}}}{(n\epsilon)^{2/7}}\right)$ and $\Tilde{O}\left(\frac{d^{1/7}\ln\frac{(n\epsilon)^2}{\beta d}}{(n\epsilon)^{2/7}}\right)$ under bounded or unbounded domain assumption, respectively (here $n$ is the sample size, $d$ is the dimension of the data, $\beta$ is the confidence level and $\epsilon$ is the private level). Then, we extend our analysis to the strongly convex case and non-smooth case (which works for generalized smooth objectives with H$\ddot{\text{o}}$lder-continuous gradients). We establish new excess risk bounds without bounded domain assumption. The results above achieve lower excess risks and gradient complexities than existing methods in their corresponding cases. Numerical experiments are conducted to justify the theoretical improvement.

FlowX: Towards Explainable Graph Neural Networks via Message Flows

We investigate the explainability of graph neural networks (GNNs) as a step towards elucidating their working mechanisms. While most current methods focus on explaining graph nodes, edges, or features, we argue that, as the inherent functional mechanism of GNNs, message flows are more natural for performing explainability. To this end, we propose a novel method here, known as FlowX, to explain GNNs by identifying important message flows. To quantify the importance of flows, we propose to follow the philosophy of Shapley values from cooperative game theory. To tackle the complexity of computing all coalitions' marginal contributions, we propose an approximation scheme to compute Shapley-like values as initial assessments of further redistribution training. We then propose a learning algorithm to train flow scores and improve explainability. Experimental studies on both synthetic and real-world datasets demonstrate that our proposed FlowX leads to improved explainability of GNNs.

Explaining the root causes of unit-level changes

Existing methods of explainable AI and interpretable ML cannot explain change in the values of an output variable for a statistical unit in terms of the change in the input values and the change in the "mechanism" (the function transforming input to output). We propose two methods based on counterfactuals for explaining unit-level changes at various input granularities using the concept of Shapley values from game theory. These methods satisfy two key axioms desirable for any unit-level change attribution method. Through simulations, we study the reliability and the scalability of the proposed methods. We get sensible results from a case study on identifying the drivers of the change in the earnings for individuals in the US.

Improving Policy Optimization with Generalist-Specialist Learning

Generalization in deep reinforcement learning over unseen environment variations usually requires policy learning over a large set of diverse training variations. We empirically observe that an agent trained on many variations (a generalist) tends to learn faster at the beginning, yet its performance plateaus at a less optimal level for a long time. In contrast, an agent trained only on a few variations (a specialist) can often achieve high returns under a limited computational budget. To have the best of both worlds, we propose a novel generalist-specialist training framework. Specifically, we first train a generalist on all environment variations; when it fails to improve, we launch a large population of specialists with weights cloned from the generalist, each trained to master a selected small subset of variations. We finally resume the training of the generalist with auxiliary rewards induced by demonstrations of all specialists. In particular, we investigate the timing to start specialist training and compare strategies to learn generalists with assistance from specialists. We show that this framework pushes the envelope of policy learning on several challenging and popular benchmarks including Procgen, Meta-World and ManiSkill.

Explainable and High-Performance Hate and Offensive Speech Detection

The spread of information through social media platforms can create environments possibly hostile to vulnerable communities and silence certain groups in society. To mitigate such instances, several models have been developed to detect hate and offensive speech. Since detecting hate and offensive speech in social media platforms could incorrectly exclude individuals from social media platforms, which can reduce trust, there is a need to create explainable and interpretable models. Thus, we build an explainable and interpretable high performance model based on the XGBoost algorithm, trained on Twitter data. For unbalanced Twitter data, XGboost outperformed the LSTM, AutoGluon, and ULMFiT models on hate speech detection with an F1 score of 0.75 compared to 0.38 and 0.37, and 0.38 respectively. When we down-sampled the data to three separate classes of approximately 5000 tweets, XGBoost performed better than LSTM, AutoGluon, and ULMFiT; with F1 scores for hate speech detection of 0.79 vs 0.69, 0.77, and 0.66 respectively. XGBoost also performed better than LSTM, AutoGluon, and ULMFiT in the down-sampled version for offensive speech detection with F1 score of 0.83 vs 0.88, 0.82, and 0.79 respectively. We use Shapley Additive Explanations (SHAP) on our XGBoost models' outputs to makes it explainable and interpretable compared to LSTM, AutoGluon and ULMFiT that are black-box models.

Adversarially Robust Learning of Real-Valued Functions

We study robustness to test-time adversarial attacks in the regression setting with $\ell_p$ losses and arbitrary perturbation sets. We address the question of which function classes are PAC learnable in this setting. We show that classes of finite fat-shattering dimension are learnable. Moreover, for convex function classes, they are even properly learnable. In contrast, some non-convex function classes provably require improper learning algorithms. We also discuss extensions to agnostic learning. Our main technique is based on a construction of an adversarially robust sample compression scheme of a size determined by the fat-shattering dimension.

RF Signal Classification with Synthetic Training Data and its Real-World Performance

Neural nets are a powerful method for the classification of radio signals in the electromagnetic spectrum. These neural nets are often trained with synthetically generated data due to the lack of diverse and plentiful real RF data. However, it is often unclear how neural nets trained on synthetic data perform in real-world applications. This paper investigates the impact of different RF signal impairments (such as phase, frequency and sample rate offsets, receiver filters, noise and channel models) modeled in synthetic training data with respect to the real-world performance. For that purpose, this paper trains neural nets with various synthetic training datasets with different signal impairments. After training, the neural nets are evaluated against real-world RF data collected by a software defined radio receiver in the field. This approach reveals which modeled signal impairments should be included in carefully designed synthetic datasets. The investigated showcase example can classify RF signals into one of 20 different radio signal types from the shortwave bands. It achieves an accuracy of up to 95 % in real-world operation by using carefully designed synthetic training data only.

Self-Healing Robust Neural Networks via Closed-Loop Control

Despite the wide applications of neural networks, there have been increasing concerns about their vulnerability issue. While numerous attack and defense techniques have been developed, this work investigates the robustness issue from a new angle: can we design a self-healing neural network that can automatically detect and fix the vulnerability issue by itself? A typical self-healing mechanism is the immune system of a human body. This biology-inspired idea has been used in many engineering designs but is rarely investigated in deep learning. This paper considers the post-training self-healing of a neural network, and proposes a closed-loop control formulation to automatically detect and fix the errors caused by various attacks or perturbations. We provide a margin-based analysis to explain how this formulation can improve the robustness of a classifier. To speed up the inference of the proposed self-healing network, we solve the control problem via improving the Pontryagin Maximum Principle-based solver. Lastly, we present an error estimation of the proposed framework for neural networks with nonlinear activation functions. We validate the performance on several network architectures against various perturbations. Since the self-healing method does not need a-priori information about data perturbations/attacks, it can handle a broad class of unforeseen perturbations.

Probabilistic PolarGMM: Unsupervised Cluster Learning of Very Noisy Projection Images of Unknown Pose

A crucial step in single particle analysis (SPA) of cryogenic electron microscopy (Cryo-EM), 2D classification and alignment takes a collection of noisy particle images to infer orientations and group similar images together. Averaging these aligned and clustered noisy images produces a set of clean images, ready for further analysis such as 3D reconstruction. Fourier-Bessel steerable principal component analysis (FBsPCA) enables an efficient, adaptable, low-rank rotation operator. We extend the FBsPCA to additionally handle translations. In this extended FBsPCA representation, we use a probabilistic polar-coordinate Gaussian mixture model to learn soft clusters in an unsupervised fashion using an expectation maximization (EM) algorithm. The obtained rotational clusters are thus additionally robust to the presence of pairwise alignment imperfections. Multiple benchmarks from simulated Cryo-EM datasets show probabilistic PolarGMM's improved performance in comparisons with standard single-particle Cryo-EM tools, EMAN2 and RELION, in terms of various clustering metrics and alignment errors.

Nonwatertight Mesh Reconstruction

Reconstructing 3D non-watertight mesh from an unoriented point cloud is an unexplored area in computer vision and computer graphics. In this project, we tried to tackle this problem by extending the learning-based watertight mesh reconstruction pipeline presented in the paper 'Shape as Points'. The core of our approach is to cast the problem as a semantic segmentation problem that identifies the region in the 3D volume where the mesh surface lies and extracts the surfaces from the detected regions. Our approach achieves compelling results compared to the baseline techniques.

Cross-Silo Federated Learning: Challenges and Opportunities

Federated learning (FL) is an emerging technology that enables the training of machine learning models from multiple clients while keeping the data distributed and private. Based on the participating clients and the model training scale, federated learning can be classified into two types: cross-device FL where clients are typically mobile devices and the client number can reach up to a scale of millions; cross-silo FL where clients are organizations or companies and the client number is usually small (e.g., within a hundred). While existing studies mainly focus on cross-device FL, this paper aims to provide an overview of the cross-silo FL. More specifically, we first discuss applications of cross-silo FL and outline its major challenges. We then provide a systematic overview of the existing approaches to the challenges in cross-silo FL by focusing on their connections and differences to cross-device FL. Finally, we discuss future directions and open issues that merit research efforts from the community.

PROTOtypical Logic Tensor Networks (PROTO-LTN) for Zero Shot Learning

Semantic image interpretation can vastly benefit from approaches that combine sub-symbolic distributed representation learning with the capability to reason at a higher level of abstraction. Logic Tensor Networks (LTNs) are a class of neuro-symbolic systems based on a differentiable, first-order logic grounded into a deep neural network. LTNs replace the classical concept of training set with a knowledge base of fuzzy logical axioms. By defining a set of differentiable operators to approximate the role of connectives, predicates, functions and quantifiers, a loss function is automatically specified so that LTNs can learn to satisfy the knowledge base. We focus here on the subsumption or \texttt{isOfClass} predicate, which is fundamental to encode most semantic image interpretation tasks. Unlike conventional LTNs, which rely on a separate predicate for each class (e.g., dog, cat), each with its own set of learnable weights, we propose a common \texttt{isOfClass} predicate, whose level of truth is a function of the distance between an object embedding and the corresponding class prototype. The PROTOtypical Logic Tensor Networks (PROTO-LTN) extend the current formulation by grounding abstract concepts as parametrized class prototypes in a high-dimensional embedding space, while reducing the number of parameters required to ground the knowledge base. We show how this architecture can be effectively trained in the few and zero-shot learning scenarios. Experiments on Generalized Zero Shot Learning benchmarks validate the proposed implementation as a competitive alternative to traditional embedding-based approaches. The proposed formulation opens up new opportunities in zero shot learning settings, as the LTN formalism allows to integrate background knowledge in the form of logical axioms to compensate for the lack of labelled examples.

Latent Augmentation Improves Graph Self-Supervised Learning

Graph self-supervised learning has been vastly employed to learn representations from unlabeled graphs. Existing methods can be roughly divided into predictive learning and contrastive learning, where the latter one attracts more research attention with better empirical performance. We argue that, however, predictive models weaponed with latent augmentations and powerful decoder could achieve comparable or even better representation power than contrastive models. In this work, we introduce data augmentations into latent space for superior generalization and better efficiency. A novel graph decoder named Wiener Graph Deconvolutional Network is correspondingly designed to perform information reconstruction from augmented latent representations. Theoretical analysis proves the superior reconstruction ability of graph wiener filter. Extensive experimental results on various datasets demonstrate the effectiveness of our approach.

Learning neural state-space models: do we need a state estimator?

In recent years, several algorithms for system identification with neural state-space models have been introduced. Most of the proposed approaches are aimed at reducing the computational complexity of the learning problem, by splitting the optimization over short sub-sequences extracted from a longer training dataset. Different sequences are then processed simultaneously within a minibatch, taking advantage of modern parallel hardware for deep learning. An issue arising in these methods is the need to assign an initial state for each of the sub-sequences, which is required to run simulations and thus to evaluate the fitting loss. In this paper, we provide insights for calibration of neural state-space training algorithms based on extensive experimentation and analyses performed on two recognized system identification benchmarks. Particular focus is given to the choice and the role of the initial state estimation. We demonstrate that advanced initial state estimation techniques are really required to achieve high performance on certain classes of dynamical systems, while for asymptotically stable ones basic procedures such as zero or random initialization already yield competitive performance.

Vision Transformer for Contrastive Clustering

Vision Transformer (ViT) has shown its advantages over the convolutional neural network (CNN) with its ability to capture global long-range dependencies for visual representation learning. Besides ViT, contrastive learning is another popular research topic recently. While previous contrastive learning works are mostly based on CNNs, some latest studies have attempted to jointly model the ViT and the contrastive learning for enhanced self-supervised learning. Despite the considerable progress, these combinations of ViT and contrastive learning mostly focus on the instance-level contrastiveness, which often overlook the contrastiveness of the global clustering structures and also lack the ability to directly learn the clustering result (e.g., for images). In view of this, this paper presents an end-to-end deep image clustering approach termed Vision Transformer for Contrastive Clustering (VTCC), which for the first time, to the best of our knowledge, unifies the Transformer and the contrastive learning for the image clustering task. Specifically, with two random augmentations performed on each image in a mini-batch, we utilize a ViT encoder with two weight-sharing views as the backbone to learn the representations for the augmented samples. To remedy the potential instability of the ViT, we incorporate a convolutional stem, which uses multiple stacked small convolutions instead of a big convolution in the patch projection layer, to split each augmented sample into a sequence of patches. With representations learned via the backbone, an instance projector and a cluster projector are further utilized for the instance-level contrastive learning and the global clustering structure learning, respectively. Extensive experiments on eight image datasets demonstrate the stability (during the training-from-scratch) and the superiority (in clustering performance) of VTCC over the state-of-the-art.

TAM: Topology-Aware Margin Loss for Class-Imbalanced Node Classification

Learning unbiased node representations under class-imbalanced graph data is challenging due to interactions between adjacent nodes. Existing studies have in common that they compensate the minor class nodes `as a group' according to their overall quantity (ignoring node connections in graph), which inevitably increase the false positive cases for major nodes. We hypothesize that the increase in these false positive cases is highly affected by the label distribution around each node and confirm it experimentally. In addition, in order to handle this issue, we propose Topology-Aware Margin (TAM) to reflect local topology on the learning objective. Our method compares the connectivity pattern of each node with the class-averaged counter-part and adaptively adjusts the margin accordingly based on that. Our method consistently exhibits superiority over the baselines on various node classification benchmark datasets with representative GNN architectures.

Video Anomaly Detection via Prediction Network with Enhanced Spatio-Temporal Memory Exchange

Video anomaly detection is a challenging task because most anomalies are scarce and non-deterministic. Many approaches investigate the reconstruction difference between normal and abnormal patterns, but neglect that anomalies do not necessarily correspond to large reconstruction errors. To address this issue, we design a Convolutional LSTM Auto-Encoder prediction framework with enhanced spatio-temporal memory exchange using bi-directionalilty and a higher-order mechanism. The bi-directional structure promotes learning the temporal regularity through forward and backward predictions. The unique higher-order mechanism further strengthens spatial information interaction between the encoder and the decoder. Considering the limited receptive fields in Convolutional LSTMs, we also introduce an attention module to highlight informative features for prediction. Anomalies are eventually identified by comparing the frames with their corresponding predictions. Evaluations on three popular benchmarks show that our framework outperforms most existing prediction-based anomaly detection methods.

Batch-Ensemble Stochastic Neural Networks for Out-of-Distribution Detection

Out-of-distribution (OOD) detection has recently received much attention from the machine learning community due to its importance in deploying machine learning models in real-world applications. In this paper we propose an uncertainty quantification approach by modelling the distribution of features. We further incorporate an efficient ensemble mechanism, namely batch-ensemble, to construct the batch-ensemble stochastic neural networks (BE-SNNs) and overcome the feature collapse problem. We compare the performance of the proposed BE-SNNs with the other state-of-the-art approaches and show that BE-SNNs yield superior performance on several OOD benchmarks, such as the Two-Moons dataset, the FashionMNIST vs MNIST dataset, FashionMNIST vs NotMNIST dataset, and the CIFAR10 vs SVHN dataset.

Noise-aware Physics-informed Machine Learning for Robust PDE Discovery

This work is concerned with discovering the governing partial differential equation (PDE) of a physical system. Existing methods have demonstrated the PDE identification from finite observations but failed to maintain satisfying performance against noisy data, partly owing to suboptimal estimated derivatives and found PDE coefficients. We address the issues by introducing a noise-aware physics-informed machine learning (nPIML) framework to discover the governing PDE from data following arbitrary distributions. Our proposals are twofold. First, we propose a couple of neural networks, namely solver and preselector, which yield an interpretable neural representation of the hidden physical constraint. After they are jointly trained, the solver network approximates potential candidates, e.g., partial derivatives, which are then fed to the sparse regression algorithm that initially unveils the most likely parsimonious PDE, decided according to the information criterion. Second, we propose the denoising physics-informed neural networks (dPINNs), based on Discrete Fourier Transform (DFT), to deliver a set of the optimal finetuned PDE coefficients respecting the noise-reduced variables. The denoising PINNs' structures are compartmentalized into forefront projection networks and a PINN, by which the formerly learned solver initializes. Our extensive experiments on five canonical PDEs affirm that the proposed framework presents a robust and interpretable approach for PDE discovery, applicable to a wide range of systems, possibly complicated by noise.

FingerGAN: A Constrained Fingerprint Generation Scheme for Latent Fingerprint Enhancement

Latent fingerprint enhancement is an essential pre-processing step for latent fingerprint identification. Most latent fingerprint enhancement methods try to restore corrupted gray ridges/valleys. In this paper, we propose a new method that formulates the latent fingerprint enhancement as a constrained fingerprint generation problem within a generative adversarial network (GAN) framework. We name the proposed network as FingerGAN. It can enforce its generated fingerprint (i.e, enhanced latent fingerprint) indistinguishable from the corresponding ground-truth instance in terms of the fingerprint skeleton map weighted by minutia locations and the orientation field regularized by the FOMFE model. Because minutia is the primary feature for fingerprint recognition and minutia can be retrieved directly from the fingerprint skeleton map, we offer a holistic framework which can perform latent fingerprint enhancement in the context of directly optimizing minutia information. This will help improve latent fingerprint identification performance significantly. Experimental results on two public latent fingerprint databases demonstrate that our method outperforms the state of the arts significantly. The codes will be available for non-commercial purposes from \url{https://github.com/HubYZ/LatentEnhancement}.

fETSmcs: Feature-based ETS model component selection

The well-developed ETS (ExponenTial Smoothing or Error, Trend, Seasonality) method incorporating a family of exponential smoothing models in state space representation has been widely used for automatic forecasting. The existing ETS method uses information criteria for model selection by choosing an optimal model with the smallest information criterion among all models fitted to a given time series. The ETS method under such a model selection scheme suffers from computational complexity when applied to large-scale time series data. To tackle this issue, we propose an efficient approach for ETS model selection by training classifiers on simulated data to predict appropriate model component forms for a given time series. We provide a simulation study to show the model selection ability of the proposed approach on simulated data. We evaluate our approach on the widely used forecasting competition data set M4, in terms of both point forecasts and prediction intervals. To demonstrate the practical value of our method, we showcase the performance improvements from our approach on a monthly hospital data set.

Data Augmentation for Dementia Detection in Spoken Language

Dementia is a growing problem as our society ages, and detection methods are often invasive and expensive. Recent deep-learning techniques can offer a faster diagnosis and have shown promising results. However, they require large amounts of labelled data which is not easily available for the task of dementia detection. One effective solution to sparse data problems is data augmentation, though the exact methods need to be selected carefully. To date, there has been no empirical study of data augmentation on Alzheimer's disease (AD) datasets for NLP and speech processing. In this work, we investigate data augmentation techniques for the task of AD detection and perform an empirical evaluation of the different approaches on two kinds of models for both the text and audio domains. We use a transformer-based model for both domains, and SVM and Random Forest models for the text and audio domains, respectively. We generate additional samples using traditional as well as deep learning based methods and show that data augmentation improves performance for both the text- and audio-based models and that such results are comparable to state-of-the-art results on the popular ADReSS set, with carefully crafted architectures and features.

Estimating Link Flows in Road Networks with Synthetic Trajectory Data Generation: Reinforcement Learning-based Approaches

This paper addresses the problem of estimating link flows in a road network by combining limited traffic volume and vehicle trajectory data. While traffic volume data from loop detectors have been the common data source for link flow estimation, the detectors only cover a subset of links. Vehicle trajectory data collected from vehicle tracking sensors are also incorporated these days. However, trajectory data are often sparse in that the observed trajectories only represent a small subset of the whole population, where the exact sampling rate is unknown and may vary over space and time. This study proposes a novel generative modelling framework, where we formulate the link-to-link movements of a vehicle as a sequential decision-making problem using the Markov Decision Process framework and train an agent to make sequential decisions to generate realistic synthetic vehicle trajectories. We use Reinforcement Learning (RL)-based methods to find the best behaviour of the agent, based on which synthetic population vehicle trajectories can be generated to estimate link flows across the whole network. To ensure the generated population vehicle trajectories are consistent with the observed traffic volume and trajectory data, two methods based on Inverse Reinforcement Learning and Constrained Reinforcement Learning are proposed. The proposed generative modelling framework solved by either of these RL-based methods is validated by solving the link flow estimation problem in a real road network. Additionally, we perform comprehensive experiments to compare the performance with two existing methods. The results show that the proposed framework has higher estimation accuracy and robustness under realistic scenarios where certain behavioural assumptions about drivers are not met or the network coverage and penetration rate of trajectory data are low.

Edge Direction-invariant Graph Neural Networks for Molecular Dipole Moments Prediction

The dipole moment is a physical quantity indicating the polarity of a molecule and is determined by reflecting the electrical properties of constituent atoms and the geometric properties of the molecule. Most embeddings used to represent graph representations in traditional graph neural network methodologies treat molecules as topological graphs, creating a significant barrier to the goal of recognizing geometric information. Unlike existing embeddings dealing with equivariance, which have been proposed to handle the 3D structure of molecules properly, our proposed embeddings directly express the physical implications of the local contribution of dipole moments. We show that the developed model works reasonably even for molecules with extended geometries and captures more interatomic interaction information, significantly improving the prediction results with accuracy comparable to ab-initio calculations.

Two-Stage Neural Contextual Bandits for Personalised News Recommendation

We consider the problem of personalised news recommendation where each user consumes news in a sequential fashion. Existing personalised news recommendation methods focus on exploiting user interests and ignores exploration in recommendation, which leads to biased feedback loops and hurt recommendation quality in the long term. We build on contextual bandits recommendation strategies which naturally address the exploitation-exploration trade-off. The main challenges are the computational efficiency for exploring the large-scale item space and utilising the deep representations with uncertainty. We propose a two-stage hierarchical topic-news deep contextual bandits framework to efficiently learn user preferences when there are many news items. We use deep learning representations for users and news, and generalise the neural upper confidence bound (UCB) policies to generalised additive UCB and bilinear UCB. Empirical results on a large-scale news recommendation dataset show that our proposed policies are efficient and outperform the baseline bandit policies.

Semantic Role Aware Correlation Transformer for Text to Video Retrieval

With the emergence of social media, voluminous video clips are uploaded every day, and retrieving the most relevant visual content with a language query becomes critical. Most approaches aim to learn a joint embedding space for plain textual and visual contents without adequately exploiting their intra-modality structures and inter-modality correlations. This paper proposes a novel transformer that explicitly disentangles the text and video into semantic roles of objects, spatial contexts and temporal contexts with an attention scheme to learn the intra- and inter-role correlations among the three roles to discover discriminative features for matching at different levels. The preliminary results on popular YouCook2 indicate that our approach surpasses a current state-of-the-art method, with a high margin in all metrics. It also overpasses two SOTA methods in terms of two metrics.

Analysis of Stochastic Processes through Replay Buffers

Replay buffers are a key component in many reinforcement learning schemes. Yet, their theoretical properties are not fully understood. In this paper we analyze a system where a stochastic process X is pushed into a replay buffer and then randomly sampled to generate a stochastic process Y from the replay buffer. We provide an analysis of the properties of the sampled process such as stationarity, Markovity and autocorrelation in terms of the properties of the original process. Our theoretical analysis sheds light on why replay buffer may be a good de-correlator. Our analysis provides theoretical tools for proving the convergence of replay buffer based algorithms which are prevalent in reinforcement learning schemes.

RoME: Role-aware Mixture-of-Expert Transformer for Text-to-Video Retrieval

Seas of videos are uploaded daily with the popularity of social channels; thus, retrieving the most related video contents with user textual queries plays a more crucial role. Most methods consider only one joint embedding space between global visual and textual features without considering the local structures of each modality. Some other approaches consider multiple embedding spaces consisting of global and local features separately, ignoring rich inter-modality correlations. We propose a novel mixture-of-expert transformer RoME that disentangles the text and the video into three levels; the roles of spatial contexts, temporal contexts, and object contexts. We utilize a transformer-based attention mechanism to fully exploit visual and text embeddings at both global and local levels with mixture-of-experts for considering inter-modalities and structures' correlations. The results indicate that our method outperforms the state-of-the-art methods on the YouCook2 and MSR-VTT datasets, given the same visual backbone without pre-training. Finally, we conducted extensive ablation studies to elucidate our design choices.

Benchmarking Bayesian Improved Surname Geocoding Against Machine Learning Methods

Bayesian Improved Surname Geocoding (BISG) is the most popular method for proxying race/ethnicity in voter registration files that do not contain it. This paper benchmarks BISG against a range of previously untested machine learning alternatives, using voter files with self-reported race/ethnicity from California, Florida, North Carolina, and Georgia. This analysis yields three key findings. First, when given the exact same inputs, BISG and machine learning perform similarly for estimating aggregate racial/ethnic composition. Second, machine learning outperforms BISG at individual classification of race/ethnicity. Third, the performance of all methods varies substantially across states. These results suggest that pre-trained machine learning models are preferable to BISG for individual classification. Furthermore, mixed results at the precinct level and across states underscore the need for researchers to empirically validate their chosen race/ethnicity proxy in their populations of interest.

Your Autoregressive Generative Model Can be Better If You Treat It as an Energy-Based One

Autoregressive generative models are commonly used, especially for those tasks involving sequential data. They have, however, been plagued by a slew of inherent flaws due to the intrinsic characteristics of chain-style conditional modeling (e.g., exposure bias or lack of long-range coherence), severely limiting their ability to model distributions properly. In this paper, we propose a unique method termed E-ARM for training autoregressive generative models that takes advantage of a well-designed energy-based learning objective. By leveraging the extra degree of freedom of the softmax operation, we are allowed to make the autoregressive model itself be an energy-based model for measuring the likelihood of input without introducing any extra parameters. Furthermore, we show that E-ARM can be trained efficiently and is capable of alleviating the exposure bias problem and increase temporal coherence for autoregressive generative models. Extensive empirical results, covering benchmarks like language modeling, neural machine translation, and image generation, demonstrate the effectiveness of the proposed approach.

Repository-Level Prompt Generation for Large Language Models of Code

With the success of large language models (LLMs) of code and their use as code assistants (e.g. Codex used in GitHub Copilot), techniques for introducing domain-specific knowledge in the prompt design process become important. In this work, we propose a framework called Repo-Level Prompt Generator that learns to generate example-specific prompts using a set of rules. These rules take context from the entire repository, thereby incorporating both the structure of the repository and the context from other relevant files (e.g. imports, parent class files). Our technique doesn't require any access to the weights of the LLM, making it applicable in cases where we only have black-box access to the LLM. We conduct experiments on the task of single-line code-autocompletion using code repositories taken from Google Code archives. We demonstrate that an oracle constructed from our proposed rules gives up to 36% relative improvement over Codex, showing the quality of the rules. Further, we show that when we train a model to select the best rule, we can achieve significant performance gains over Codex. The code for our work can be found at: https://github.com/shrivastavadisha/repo_level_prompt_generation.

Prediction Errors for Penalized Regressions based on Generalized Approximate Message Passing

We discuss the prediction accuracy of assumed statistical models in terms of prediction errors for the generalized linear model and penalized maximum likelihood methods. We derive the forms of estimators for the prediction errors: $C_p$ criterion, information criteria, and leave-one-out cross validation (LOOCV) error, using the generalized approximate message passing (GAMP) algorithm and replica method. These estimators coincide with each other when the number of model parameters is sufficiently small; however, there is a discrepancy between them in particular in the overparametrized region where the number of model parameters is larger than the data dimension. In this paper, we review the prediction errors and corresponding estimators, and discuss their differences. In the framework of GAMP, we show that the information criteria can be expressed by using the variance of the estimates. Further, we demonstrate how to approach LOOCV error from the information criteria by utilizing the expression provided by GAMP.

On Comparison of Encoders for Attention based End to End Speech Recognition in Standalone and Rescoring Mode

The streaming automatic speech recognition (ASR) models are more popular and suitable for voice-based applications. However, non-streaming models provide better performance as they look at the entire audio context. To leverage the benefits of the non-streaming model in streaming applications like voice search, it is commonly used in second pass re-scoring mode. The candidate hypothesis generated using steaming models is re-scored using a non-streaming model. In this work, we evaluate the non-streaming attention-based end-to-end ASR models on the Flipkart voice search task in both standalone and re-scoring modes. These models are based on Listen-Attend-Spell (LAS) encoder-decoder architecture. We experiment with different encoder variations based on LSTM, Transformer, and Conformer. We compare the latency requirements of these models along with their performance. Overall we show that the Transformer model offers acceptable WER with the lowest latency requirements. We report a relative WER improvement of around 16% with the second pass LAS re-scoring with latency overhead under 5ms. We also highlight the importance of CNN front-end with Transformer architecture to achieve comparable word error rates (WER). Moreover, we observe that in the second pass re-scoring mode all the encoders provide similar benefits whereas the difference in performance is prominent in standalone text generation mode.

Breast Cancer Classification using Deep Learned Features Boosted with Handcrafted Features

Breast cancer is one of the leading causes of death among women across the globe. It is difficult to treat if detected at advanced stages, however, early detection can significantly increase chances of survival and improves lives of millions of women. Given the widespread prevalence of breast cancer, it is of utmost importance for the research community to come up with the framework for early detection, classification and diagnosis. Artificial intelligence research community in coordination with medical practitioners are developing such frameworks to automate the task of detection. With the surge in research activities coupled with availability of large datasets and enhanced computational powers, it expected that AI framework results will help even more clinicians in making correct predictions. In this article, a novel framework for classification of breast cancer using mammograms is proposed. The proposed framework combines robust features extracted from novel Convolutional Neural Network (CNN) features with handcrafted features including HOG (Histogram of Oriented Gradients) and LBP (Local Binary Pattern). The obtained results on CBIS-DDSM dataset exceed state of the art.

Memory-Guided Multi-View Multi-Domain Fake News Detection

The wide spread of fake news is increasingly threatening both individuals and society. Great efforts have been made for automatic fake news detection on a single domain (e.g., politics). However, correlations exist commonly across multiple news domains, and thus it is promising to simultaneously detect fake news of multiple domains. Based on our analysis, we pose two challenges in multi-domain fake news detection: 1) domain shift, caused by the discrepancy among domains in terms of words, emotions, styles, etc. 2) domain labeling incompleteness, stemming from the real-world categorization that only outputs one single domain label, regardless of topic diversity of a news piece. In this paper, we propose a Memory-guided Multi-view Multi-domain Fake News Detection Framework (M$^3$FEND) to address these two challenges. We model news pieces from a multi-view perspective, including semantics, emotion, and style. Specifically, we propose a Domain Memory Bank to enrich domain information which could discover potential domain labels based on seen news pieces and model domain characteristics. Then, with enriched domain information as input, a Domain Adapter could adaptively aggregate discriminative information from multiple views for news in various domains. Extensive offline experiments on English and Chinese datasets demonstrate the effectiveness of M$^3$FEND, and online tests verify its superiority in practice. Our code is available at https://github.com/ICTMCG/M3FEND.

Bounding the Width of Neural Networks via Coupled Initialization -- A Worst Case Analysis

A common method in training neural networks is to initialize all the weights to be independent Gaussian vectors. We observe that by instead initializing the weights into independent pairs, where each pair consists of two identical Gaussian vectors, we can significantly improve the convergence analysis. While a similar technique has been studied for random inputs [Daniely, NeurIPS 2020], it has not been analyzed with arbitrary inputs. Using this technique, we show how to significantly reduce the number of neurons required for two-layer ReLU networks, both in the under-parameterized setting with logistic loss, from roughly $\gamma^{-8}$ [Ji and Telgarsky, ICLR 2020] to $\gamma^{-2}$, where $\gamma$ denotes the separation margin with a Neural Tangent Kernel, as well as in the over-parameterized setting with squared loss, from roughly $n^4$ [Song and Yang, 2019] to $n^2$, implicitly also improving the recent running time bound of [Brand, Peng, Song and Weinstein, ITCS 2021]. For the under-parameterized setting we also prove new lower bounds that improve upon prior work, and that under certain assumptions, are best possible.

Structural Entropy Guided Graph Hierarchical Pooling

Following the success of convolution on non-Euclidean space, the corresponding pooling approaches have also been validated on various tasks regarding graphs. However, because of the fixed compression quota and stepwise pooling design, these hierarchical pooling methods still suffer from local structure damage and suboptimal problem. In this work, inspired by structural entropy, we propose a hierarchical pooling approach, SEP, to tackle the two issues. Specifically, without assigning the layer-specific compression quota, a global optimization algorithm is designed to generate the cluster assignment matrices for pooling at once. Then, we present an illustration of the local structure damage from previous methods in the reconstruction of ring and grid synthetic graphs. In addition to SEP, we further design two classification models, SEP-G and SEP-N for graph classification and node classification, respectively. The results show that SEP outperforms state-of-the-art graph pooling methods on graph classification benchmarks and obtains superior performance on node classifications.

Class Impression for Data-free Incremental Learning

Standard deep learning-based classification approaches require collecting all samples from all classes in advance and are trained offline. This paradigm may not be practical in real-world clinical applications, where new classes are incrementally introduced through the addition of new data. Class incremental learning is a strategy allowing learning from such data. However, a major challenge is catastrophic forgetting, i.e., performance degradation on previous classes when adapting a trained model to new data. Prior methodologies to alleviate this challenge save a portion of training data require perpetual storage of such data that may introduce privacy issues. Here, we propose a novel data-free class incremental learning framework that first synthesizes data from the model trained on previous classes to generate a \ours. Subsequently, it updates the model by combining the synthesized data with new class data. Furthermore, we incorporate a cosine normalized Cross-entropy loss to mitigate the adverse effects of the imbalance, a margin loss to increase separation among previous classes and new ones, and an intra-domain contrastive loss to generalize the model trained on the synthesized data to real data. We compare our proposed framework with state-of-the-art methods in class incremental learning, where we demonstrate improvement in accuracy for the classification of 11,062 echocardiography cine series of patients.

Transferring Fairness under Distribution Shifts via Fair Consistency Regularization

The increasing reliance on ML models in high-stakes tasks has raised a major concern on fairness violations. Although there has been a surge of work that improves algorithmic fairness, most of them are under the assumption of an identical training and test distribution. In many real-world applications, however, such an assumption is often violated as previously trained fair models are often deployed in a different environment, and the fairness of such models has been observed to collapse. In this paper, we study how to transfer model fairness under distribution shifts, a widespread issue in practice. We conduct a fine-grained analysis of how the fair model is affected under different types of distribution shifts and find that domain shifts are more challenging than subpopulation shifts. Inspired by the success of self-training in transferring accuracy under domain shifts, we derive a sufficient condition for transferring group fairness. Guided by it, we propose a practical algorithm with a fair consistency regularization as the key component. A synthetic dataset benchmark, which covers all types of distribution shifts, is deployed for experimental verification of the theoretical findings. Experiments on synthetic and real datasets including image and tabular data demonstrate that our approach effectively transfers fairness and accuracy under various distribution shifts.

APPFLChain: A Privacy Protection Distributed Artificial-Intelligence Architecture Based on Federated Learning and Consortium Blockchain

Recent research in Internet of things has been widely applied for industrial practices, fostering the exponential growth of data and connected devices. Henceforth, data-driven AI models would be accessed by different parties through certain data-sharing policies. However, most of the current training procedures rely on the centralized data-collection strategy and a single computational server. However, such a centralized scheme may lead to many issues. Customer data stored in a centralized database may be tampered with so the provenance and authenticity of data cannot be justified. Once the aforementioned security concerns occur, the credibility of the trained AI models would be questionable and even unfavorable outcomes might be produced at the test stage. Lately, blockchain and AI, the two core technologies in Industry 4.0 and Web 3.0, have been explored to facilitate the decentralized AI training strategy. To serve on this very purpose, we propose a new system architecture called APPFLChain, namely an integrated architecture of a Hyperledger Fabric-based blockchain and a federated-learning paradigm. Our proposed new system allows different parties to jointly train AI models and their customers or stakeholders are connected by a consortium blockchain-based network. Our new system can maintain a high degree of security and privacy as users do not need to share sensitive personal information to the server. For numerical evaluation, we simulate a real-world scenario to illustrate the whole operational process of APPFLChain. Simulation results show that taking advantage of the characteristics of consortium blockchain and federated learning, APPFLChain can demonstrate favorable properties including untamperability, traceability, privacy protection, and reliable decision-making.

Malware Detection and Prevention using Artificial Intelligence Techniques

With the rapid technological advancement, security has become a major issue due to the increase in malware activity that poses a serious threat to the security and safety of both computer systems and stakeholders. To maintain stakeholders, particularly, end users security, protecting the data from fraudulent efforts is one of the most pressing concerns. A set of malicious programming code, scripts, active content, or intrusive software that is designed to destroy intended computer systems and programs or mobile and web applications is referred to as malware. According to a study, naive users are unable to distinguish between malicious and benign applications. Thus, computer systems and mobile applications should be designed to detect malicious activities towards protecting the stakeholders. A number of algorithms are available to detect malware activities by utilizing novel concepts including Artificial Intelligence, Machine Learning, and Deep Learning. In this study, we emphasize Artificial Intelligence (AI) based techniques for detecting and preventing malware activity. We present a detailed review of current malware detection technologies, their shortcomings, and ways to improve efficiency. Our study shows that adopting futuristic approaches for the development of malware detection applications shall provide significant advantages. The comprehension of this synthesis shall help researchers for further research on malware detection and prevention using AI.

Generalized Beliefs for Cooperative AI

Self-play is a common paradigm for constructing solutions in Markov games that can yield optimal policies in collaborative settings. However, these policies often adopt highly-specialized conventions that make playing with a novel partner difficult. To address this, recent approaches rely on encoding symmetry and convention-awareness into policy training, but these require strong environmental assumptions and can complicate policy training. We therefore propose moving the learning of conventions to the belief space. Specifically, we propose a belief learning model that can maintain beliefs over rollouts of policies not seen at training time, and can thus decode and adapt to novel conventions at test time. We show how to leverage this model for both search and training of a best response over various pools of policies to greatly improve ad-hoc teamplay. We also show how our setup promotes explainability and interpretability of nuanced agent conventions.

Predicting the Need for Blood Transfusion in Intensive Care Units with Reinforcement Learning

As critically ill patients frequently develop anemia or coagulopathy, transfusion of blood products is a frequent intervention in the Intensive Care Units (ICU). However, inappropriate transfusion decisions made by physicians are often associated with increased risk of complications and higher hospital costs. In this work, we aim to develop a decision support tool that uses available patient information for transfusion decision-making on three common blood products (red blood cells, platelets, and fresh frozen plasma). To this end, we adopt an off-policy batch reinforcement learning (RL) algorithm, namely, discretized Batch Constrained Q-learning, to determine the best action (transfusion or not) given observed patient trajectories. Simultaneously, we consider different state representation approaches and reward design mechanisms to evaluate their impacts on policy learning. Experiments are conducted on two real-world critical care datasets: the MIMIC-III and the UCSF. Results demonstrate that policy recommendations on transfusion achieved comparable matching against true hospital policies via accuracy and weighted importance sampling evaluations on the MIMIC-III dataset. Furthermore, a combination of transfer learning (TL) and RL on the data-scarce UCSF dataset can provide up to $17.02% improvement in terms of accuracy, and up to 18.94% and 21.63% improvement in jump-start and asymptotic performance in terms of weighted importance sampling averaged over three transfusion tasks. Finally, simulations on transfusion decisions suggest that the transferred RL policy could reduce patients' estimated 28-day mortality rate by 2.74% and decreased acuity rate by 1.18% on the UCSF dataset.

Spatiotemporal Data Mining: A Survey

Spatiotemporal data mining aims to discover interesting, useful but non-trivial patterns in big spatial and spatiotemporal data. They are used in various application domains such as public safety, ecology, epidemiology, earth science, etc. This problem is challenging because of the high societal cost of spurious patterns and exorbitant computational cost. Recent surveys of spatiotemporal data mining need update due to rapid growth. In addition, they did not adequately survey parallel techniques for spatiotemporal data mining. This paper provides a more up-to-date survey of spatiotemporal data mining methods. Furthermore, it has a detailed survey of parallel formulations of spatiotemporal data mining.

Design and Analysis of Robust Resilient Diffusion over Multi-Task Networks Against Byzantine Attacks

This paper studies distributed diffusion adaptation over clustered multi-task networks in the presence of impulsive interferences and Byzantine attacks. We develop a robust resilient diffusion least mean Geman-McClure-estimation (RDLMG) algorithm based on the cost function used by the Geman-McClure estimator, which can reduce the sensitivity to large outliers and make the algorithm robust under impulsive interferences. Moreover, the mean sub-sequence reduced method, in which each node discards the extreme value information of cost contributions received from its neighbors, can make the network resilient against Byzantine attacks. In this regard, the proposed RDLMG algorithm ensures that all normal nodes converge to their ideal states with cooperation among nodes. A statistical analysis of the RDLMG algorithm is also carried out in terms of mean and mean-square performances. Numerical results evaluate the proposed RDLMG algorithm in applications to multi-target localization and multi-task spectrum sensing.

HyGNN: Drug-Drug Interaction Prediction via Hypergraph Neural Network

Drug-Drug Interactions (DDIs) may hamper the functionalities of drugs, and in the worst scenario, they may lead to adverse drug reactions (ADRs). Predicting all DDIs is a challenging and critical problem. Most existing computational models integrate drug-centric information from different sources and leverage them as features in machine learning classifiers to predict DDIs. However, these models have a high chance of failure, especially for the new drugs when all the information is not available. This paper proposes a novel Hypergraph Neural Network (HyGNN) model based on only the SMILES string of drugs, available for any drug, for the DDI prediction problem. To capture the drug similarities, we create a hypergraph from drugs' chemical substructures extracted from the SMILES strings. Then, we develop HyGNN consisting of a novel attention-based hypergraph edge encoder to get the representation of drugs as hyperedges and a decoder to predict the interactions between drug pairs. Furthermore, we conduct extensive experiments to evaluate our model and compare it with several state-of-the-art methods. Experimental results demonstrate that our proposed HyGNN model effectively predicts DDIs and impressively outperforms the baselines with a maximum ROC-AUC and PR-AUC of 97.9% and 98.1%, respectively.

Modeling Oceanic Variables with Dynamic Graph Neural Networks

Researchers typically resort to numerical methods to understand and predict ocean dynamics, a key task in mastering environmental phenomena. Such methods may not be suitable in scenarios where the topographic map is complex, knowledge about the underlying processes is incomplete, or the application is time critical. On the other hand, if ocean dynamics are observed, they can be exploited by recent machine learning methods. In this paper we describe a data-driven method to predict environmental variables such as current velocity and sea surface height in the region of Santos-Sao Vicente-Bertioga Estuarine System in the southeastern coast of Brazil. Our model exploits both temporal and spatial inductive biases by joining state-of-the-art sequence models (LSTM and Transformers) and relational models (Graph Neural Networks) in an end-to-end framework that learns both the temporal features and the spatial relationship shared among observation sites. We compare our results with the Santos Operational Forecasting System (SOFS). Experiments show that better results are attained by our model, while maintaining flexibility and little domain knowledge dependency.

On how to avoid exacerbating spurious correlations when models are overparameterized

Overparameterized models fail to generalize well in the presence of data imbalance even when combined with traditional techniques for mitigating imbalances. This paper focuses on imbalanced classification datasets, in which a small subset of the population -- a minority -- may contain features that correlate spuriously with the class label. For a parametric family of cross-entropy loss modifications and a representative Gaussian mixture model, we derive non-asymptotic generalization bounds on the worst-group error that shed light on the role of different hyper-parameters. Specifically, we prove that, when appropriately tuned, the recently proposed VS-loss learns a model that is fair towards minorities even when spurious features are strong. On the other hand, alternative heuristics, such as the weighted CE and the LA-loss, can fail dramatically. Compared to previous works, our bounds hold for more general models, they are non-asymptotic, and, they apply even at scenarios of extreme imbalance.

Self-Supervised 3D Monocular Object Detection by Recycling Bounding Boxes

Modern object detection architectures are moving towards employing self-supervised learning (SSL) to improve performance detection with related pretext tasks. Pretext tasks for monocular 3D object detection have not yet been explored yet in literature. The paper studies the application of established self-supervised bounding box recycling by labeling random windows as the pretext task. The classifier head of the 3D detector is trained to classify random windows containing different proportions of the ground truth objects, thus handling the foreground-background imbalance. We evaluate the pretext task using the RTM3D detection model as baseline, with and without the application of data augmentation. We demonstrate improvements of between 2-3 % in mAP 3D and 0.9-1.5 % BEV scores using SSL over the baseline scores. We propose the inverse class frequency re-weighted (ICFW) mAP score that highlights improvements in detection for low frequency classes in a class imbalanced dataset with long tails. We demonstrate improvements in ICFW both mAP 3D and BEV scores to take into account the class imbalance in the KITTI validation dataset. We see 4-5 % increase in ICFW metric with the pretext task.