
Deep Dictionary Learning
Description
Deep Dictionary Learning: Synthesis and Analysis challenges the way we think about deep learning. Instead of treating models as opaque stacks of nonlinear layers, this book reveals a cleaner, more structured alternative rooted in representation and insight. It reframes "depth" not as complexity for its own sake, but as a principled way to build meaning from data-offering a perspective that is both powerful and refreshingly transparent.
Cutting through the noise of black-box models, this book brings clarity to modern machine learning. It unifies ideas that are usually scattered across the literature and shows how they come together in a single, elegant framework. Through diverse applications-from imaging to data analysis-it demonstrates how these methods can be used in practice while retaining interpretability and control. Readers will come away not just with techniques, but with a sharper way of thinking about learning systems.
This book is for readers who are not satisfied with simply using deep learning, but want to truly understand it. Ideal for graduate students, researchers, and practitioners in machine learning, data science, and signal processing, it speaks to those who question black-box approaches and are looking for models that are both powerful and explainable.
More details
Other editions
Additional editions
Person
Angshul Majumdar is a Professor at the Indraprastha Institute of Information Technology Delhi, where he has been a faculty member since 2012. He received his MASc (2009) and PhD (2012) in Electrical and Computer Engineering from the University of British Columbia, Vancouver. His research interests lie in signal processing and machine learning, with a focus on sparse and low-rank modelling, dictionary and transform learning, and their applications in imaging and data analytics. He has co-authored over 200 papers in journals and top-tier conferences, written three books, and co-edited two more. He also holds ten US and European patents. Angshul currently serves as an Associate Editor for IEEE Transactions on Multimedia and as a Senior Area Editor for both the IEEE Open Journal of Signal Processing and IEEE Signal Processing Letters.
Content
Foreword
Acknowledgements
1 Introduction, Motivation and Background
2 Deep Dictionary & Transform Learning: Greedy vs End-to-End
3 Inverse Problems
4 Supervised Learning
5 Clustering
6 Domain Adaptation
7 Convolutional Models
8 Conclusion