Bit of a Tangent

012 | How Deep Learning Does Magic

August 26, 2019

This is a discussion about why deep neural nets are unreasonably effective. Gianluca and Jared examine the relationships between neural architectures and the laws of physics that govern our Universe—exploring brains, human language, and linear functions. Nothing could have prepared them for the territories this episode expanded to, so strap yourself in!



AlphaGo beating Lee Sedol at Go:

OpenAI Five:

Taylor series/expansions video from 3Blue1Brown:

Physicist Max Tegmark:

Tegmark’s great talk on connections between physics and deep learning (which formed much of the inspiration for this conversation):

Universal Approximation Theorem:

A refresher on “Map vs. Territory”:

Ada Lovelace (who worked on Babbage’s Analytical Engine):

Manifolds and their topology:

Binary trees:

Markov process:

OpenAIs GPT-2:

Play with GPT-2 in your browser here:

Lex Fridman’s MIT Artificial Intelligence podcast:

The Scientific Odyssey podcast:

Podbean App

Play this podcast on Podbean App