2

I'm looking for a modern machine learning book with graduate-level treatment of more recent topics such as diffusion and generative models, transformers etc.

I have a hard copy of Deep Learning by Goodfellow and Bengio; while I liked the book and read it extensively when it was published, it is a bit dated now.

I'm considering Probabilistic Machine Learning: Advanced Topics by Kevin Patrick Murphy. But maybe there are better alternatives.

I need this for my comprehensive Ph.D. examination. Any suggestions are very appreciated.

P.S. I'm also aware of this stack post, but the list of references there is quite dated or introductory.

user74376
  • 121
  • 3

2 Answers2

3

Berkeley CS294-158 is a graduate-level course on deep unsupervised learning. They cover a lot of architectures used in modern generative modeling. They have recorded lectures and slides online.

Stanford CS236 is a course on deep generative modeling. They don't have a textbook, but they have course notes and slides online.

Alexander Wan
  • 1,409
  • 1
  • 12
2

Here are my suggestions:

  1. Deep Learning: Foundations and Concepts (2024) by Bishop – A comprehensive and up-to-date text covering fundamental and advanced topics in deep learning.
  2. Reinforcement Learning: An Introduction by Sutton – A classic and authoritative reference on reinforcement learning, perfect for understanding RL fundamentals and recent advances.
  3. Multi-Agent Reinforcement Learning: Foundations and Modern Approaches (2024) – Focuses on the growing field of multi-agent systems, with a solid theoretical foundation and modern applications.
  4. The Principles of Deep Learning Theory – Offers a theoretical perspective on deep learning, bridging the gap between statistical physics and machine learning.
  5. Foundations of Computer Vision (2024) – A modern take on computer vision, providing both foundational knowledge and cutting-edge techniques.

Also, don't forget the famous list of 27 papers (reportedly shared by Ilya Sutskever with John Carmack). As Sutskever mentioned:

"If you truly master these, you’ll understand 90% of what matters in modern AI."

ofou
  • 201
  • 1
  • 4