Skip to content
Hallway
0:00 / 0:00

#maxtegmark#deeplearning#physics

Connections between physics and deep learning

MITCBMM

62K subscribers • 1 video on Hallway

YouTube
  • Recommended by 3 people: 3 on Hacker News.
  • Presented by Max Tegmark, Professor of Physics at MIT
  • May be out of date. Uploaded 10 years ago, in an area where tools change fast.

Advanced. Deep dive for people who know the field.

66K views10 years ago3 recommendations

AI and machine learningComputer scienceNeural networks and deep learningMath for machine learningComputation and complexity

Max Tegmark explains why deep learning works so well by connecting the laws of physics and symmetries in physical data to neural network structures.

Max Tegmark - MIT

Why it's here

3 people recommend this

What people say. Recommended as a fascinating talk explaining the mathematical and physical reasons behind the success of deep neural networks.

tedsanderson Hacker News • Recommends

On “Why does deep and cheap learning work so well?”

Here's a related 2016 talk by Max Tegmark (second author) on connections between deep learning and physics: link to this video The gist of it is that physical data tends to have symmetries, and these symmetries make descriptions of the data very compressible into relatively small neural circuits. Random data does not have this property, and cannot be learned easily. Super fascinating.
View on Hacker News

adityabon Hacker News • Recommends

On “Link Between Deep Neural Networks and the Nature of the Universe”

…he intellect. On a dreamy note, I wonder if it would be possible to draw Category-theoretic parallels between some physical theories and statistical learning theory. There is so much to learn, and I am trying my best to teach myself (on the side) the mathematics that they don't teach in my CS grad school. So much to learn, such little time. :) 1 link to this video 2 link…
View on Hacker News

jonmc12on Hacker News • Recommends

On “Game Theory Reveals the Future of Deep Learning”

I think the argument for using physics models for constraints in a DL system was a more clear argument link to this video
View on Hacker News

clickokon Hacker News • Mentions

On “iGAN – Deep learning software that generates images with a few brushstrokes”

…t one, and slightly less likely the one after that, whereas in a Markovian setting my confidence in my predictions should decay much more quickly. So in answer to the grandparent, such a thing should be reasonably straightforward to build if it doesn't exist already, and it may offer improvements over a similar model based on Markov chains. --- 0. link to this video 1. link 2. Why is this? Lin & Tegmark offer details in the paper, but it comes from the fact that the singular values of the transition matrix are all less than or equal to one (an aperiodic & ergodic transition matrix has only one singular value equal to one), and so the other singular vectors fall away exponentially quickly, with the ex…
View on Hacker News

mikhailfrancoon Hacker News • Mentions

On “Deep Learning and Quantum Entanglement [pdf]”

…tmax/sigmoid logistic classifiers). An even more obscure tangent is that most of the financial market is now constructed from 'options', which track continuous values with a continuous payoff, but offer a choice of execution to collapse the option and create a real financial result, e.g. see N.N.Taleb's dicussion of optionality 5. 1 link 2 link to this video 3 link 4 link 5 link…
View on Hacker News

krackerson Hacker News • Mentions

On “How LLMs Work”

See Tegmark's "why does deep cheap learning work so well" (well not so cheap anymore...) link to this video
View on Hacker News