Part 3: Research + Resources

Conferences

Interestingly, the machine learning community does not actively publish in academic journals. The pace of research is simply too fast for publishers to keep up. Instead, the majority of active ML research can be found published in conference proceedings, or even directly on the ArXiv. While there are many ML conferences, two that you should know out of the gate are:

1. Neural Information Processing (NeurIPS)
2. International Conference on Machine Learning (ICML)

Importantly, these machine learning conferences are peer-reviewed, and have extraordinarily low acceptance rates (<10%). Therefore, if you're reading a paper that comes from NeurIPS or ICML, odds are, its a very good one. You can find all papers + their corresponding peer reviews on OpenReview.

Using Google Scholar

Be sure to use Google Scholar to search for this material. In Settings, you can set up a library link to your UMD account which should get you past most paywalls. If not, you can use the "Reload at UMD" button which should solve the other cases.

Papers

As before, good conference / journal papers can be an excellent place to start. I recommend looking though the best or outstanding papers at these forums which can often be found on the archived conference website (example).

One particularly strong paper that captures many of the important themes of our group's ML interest is: https://arxiv.org/pdf/2112.03235

Dissertations

• Patrick Kidger
• Miles Cranmer
• Matthew Tannick

While excellent astrodynamics dissertations can come from anywhere, a good recommendation is to consult recent dissertations from Stanford, MIT, Carnegie Mellon all are excellent places to begin learning.

Textbooks

• Dive into Deep Learning ( )
• Algorithms for Decision Making - Mykel Kochenderfer ( )
• The Little Book of Deep Learning ( )
• Deep Learning -- Goodfellow ( )
• RL: An Introduction -- Sutton and Barto ( )
• Probabalistic Machine Learning, Murphy ( )
• Data Driven Science & Engineering Machine Learning, Dynamical Systems, and Control -- Brunton + Kutz (?)

Additional Resources

• Grant Sanderson's 3Blue1Brown Neural Network Series ( )
• Andrew Ng' Machine Learning Course ( )
• David Silver's Reinforcement Learning Course ( )
• Steve Brunton's Physics Informed Machine Learning ( )
• Spinning up with DRL ( )
• Tensorflow Playground ( )
• Alpha Go Movie ( )
• Andrej Karpathy Neural Networks: Zero to Hero ( -- Advanced)

Machine Learning Concepts to Learn

See Patrick Kidger's "Just Know Stuff".