Physics and Machine Learning
The conversation explores the surprising effectiveness of stochastic gradient descent and the potential of physics-inspired models in machine learning. Joscha highlights how physicists often rely on differential equations to describe systems, which can be beneficial but may overlook the discrete nature of many machine learning problems. This raises intriguing questions about whether machine learning should be built on discrete automata rather than traditional linear algebra approaches.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Joscha Bach - Why Your Thoughts Aren't Yours.
Related Questions