Information, Physics, and Machine Learning

Zohar Nussinov from the Department of Physics, Washington University in St. Louis, host this Saturday Science Lecture on April 9, 2022. We briefly review how information content (and its average value, the entropy) plays a key role in various branches of physics (from idealized engines to black holes) and illustrate the use of entropy in solving various known logic puzzles, reconstructing redacted messages, and other tasks. We conclude by, very qualitatively, sketching several key physics based ideas in artificial neural networks and explain why entropy based measures underlie many current machine learning endeavors.