Andrew Dudzik - Three Problems in the Mathematics of Deep Learning - YouTube
Follow along using the transcript. Follow along using the transcript. New 3BLUE1BROWN SERIES S4 E6 New Sponsored Sponsored Sponsored Sponsored @AustenAlexanderTV
Neural networks, particularly LLMs, are notoriously poor at algorithmic tasks, such as sorting, shortest path, and even basic arithmetic. Across three papers, we explored the problem of "aligning" architectures to classical computer programs, and showed that this question relates to familiar mathematical concepts: polynomial functors, cohomology, and higher categories. Andrew Dudzik (Google DeepMind) === Find this and many more scientific videos on https://www.carmin.tv/ - a French video platform for mathematics and their interactions with other sciences offering extra functionalities…
saved by
related reading
- Neural Networks, Manifolds, and Topology -- colah's blogcolah.github.io
- A Theory of Deep Learning | Elements of a Vector Spaceelonlit.com
- Deep Dive into LLMs like ChatGPTyoutube.com
- Neural Networks, Types, and Functional Programming -- colah's blogcolah.github.io
- Maybe I was too harsh on deep learning theory (three days ago) — LessWronglesswrong.com
- We need to talk about AI in mathematicsyoutube.com
- Deep learning as program synthesis — LessWronglesswrong.com
- Reiner Pope – The math behind how LLMs are trained and serveddwarkesh.com
- ICLR 2021 Keynote - "Geometric Deep Learning: The Erlangen Programme of ML" - M Bronsteinyoutube.com
- [2604.21691] There Will Be a Scientific Theory of Deep Learningarxiv.org
- On Developing a Mathematical Theory of Interpretability — LessWronglesswrong.com
- [2604.07242] Weaves, Wires, and Morphisms: Formalizing and Implementing the Algebra of Deep Learningarxiv.org