| @misc{duncan2024uann, | |
| title={Universal Adaptive Neural Network: A Multi-Modal Mixture of Experts Model for Adaptive AI Agents}, | |
| author={Dennis Duncan}, | |
| year={2024}, | |
| howpublished={\url{https://huggingface.co/dnnsdunca/UANN}}, | |
| note={Hugging Face Transformers Library}, | |
| abstract={ | |
| This repository presents the Universal Adaptive Neural Network (UANN), a versatile neural network architecture designed for multi-modal AI agents. The UANN integrates a Mixture of Experts (MoE) model with a primary gating network and specialized secondary experts. The primary agent dynamically controls the gating mechanism, allowing the model to adaptively allocate computational resources across various modalities, including vision, audio, and sensor data. This architecture enables efficient processing and high adaptability in complex environments, making it suitable for advanced AI applications in robotics, game AI, and more. The total number of parameters in the model is approximately 529,411. | |
| }, | |
| datasets={CIFAR-10, LibriSpeech, UCI HAR Dataset}, | |
| keywords={Universal Adaptive Neural Network, UANN, Mixture of Experts, MoE, AI agents, multi-modal learning, neural network architecture, adaptive learning, vision, audio, sensors} | |
| } | |