Applied Deep Learning on Graphs

With their combined expertise spanning cutting-edge AI product development at industry giants such as Walmart, Adobe, Samsung, and Arista Networks, Lakshya and Subhajoy provide real-world insights into the transformative world of graph neural networks (GNNs). This book demystifies GNNs, guiding you from foundational concepts to advanced techniques and real-world applications. Youll see how graph data structures power todays interconnected world, why specialized deep learning approaches are essential, and how to address challenges with existing methods. Youll start by dissecting early graph representation techniques such as DeepWalk and node2vec. From there, the book takes you through popular GNN architectures, covering graph convolutional and attention networks, autoencoder models, LLMs, and technologies such as retrieval augmented generation on graph data. With a strong theoretical grounding, youll seamlessly navigate practical implementations, mastering the critical topics of scalability, interpretability, and application domains such as NLP, recommendations, and computer vision. By the end of this book, youll have mastered the underlying ideas and practical coding skills needed to innovate beyond current methods and gained strategic insights into the future of GNN technologies.