Get started
Work

Les Houches Lectures on Deep Learning at Large & Infinite Width

Year: 2023
Type: preprint
Abstract: These lectures, presented at the 2022 Les Houches Summer School on Statistical Physics and Machine Learning, focus on the infinite-width limit and large-width regime of deep neural networks. Topics covered include various statistical and dynamical properties of these networks. In particular, the lecturers discuss properties of random deep neural networks; connections between trained deep neural networks, linear models, kernels, and Gaussian processes that arise in the infinite-width limit; and perturbative and non-perturbative treatments of large but finite-width networks, at initialization and after training.
DOI: 10.48550/arxiv.2309.01592
Authors: Bahri, Yasaman, Hanin, Boris, Brossollet, Antonin, Erba, Vittorio, Keup, Christian, Pacelli, Rosalba, Simon, James B.
Institutions: Google (United States), Princeton University, École Normale Supérieure - PSL, Laboratoire de Physique de l'Ecole Normale Supérieure, École Polytechnique Fédérale de Lausanne, Politecnico di Torino, Bocconi University, University of California, Berkeley
Language: English
Open Access status: green
OpenAlex ID: W4386556769
Cites:
Cited by:
Related to: 10