Overview

Frontiers in Deep Learning: Advanced Models, Training Paradigms, and Open Problems presents a comprehensive exploration of emerging directions in deep learning beyond traditional architectures and training methods. The book critically examines the limitations of backpropagation, biological implausibility, memory inefficiency, and catastrophic forgetting, while introducing innovative alternatives such as spiking neural networks, predictive coding and equilibrium propagation. It covers advanced topics like meta-learning, deep equilibrium models, transformer architectures, graph neural networks, neuro-symbolic AI, self-supervised learning, diffusion models, scalable training strategies, and efficient inference techniques. The work emphasizes causal learning, adversarial robustness, uncertainty quantification, explainable AI, and multi-modal learning as essential components for trustworthy and generalizable AI systems. By bridging theoretical foundations with real-world applications in healthcare, scientific discovery, and automation, the book provides a forward-looking vision of deep learning that moves toward more adaptive, interpretable, and energy-efficient artificial intelligence.

Full Product Details

9786209397936

Table of Contents

Reviews

Author Information

Tab Content 6

Countries Available