Nonlinear Physics

Name	Category	Phone	Office
Kolinski John	Post-Docs	02 6586 104	Levin 17
Meerson Baruch	Faculty	02 6584 470	Levin 18
Moshe Michael	Faculty	02 5494 819	Kaplun 102
Palacio Jorge	Post-Docs	02 6585 707	Levin 14
Rokni Mordechai	Emeriti	02 6584 497	Danciger B 119
Sahaf Michal	Graduate Students	02 6586 104	Levin 22
Schwob Yehuda	Emeriti	02 6584 331	Kaplun 33
Sharon Eran	Faculty	02 6585 674	Levin 16b
Smith Naftali	Graduate Students	02 6585 263	Levin 14
Vilenkin Arkady	Senior Scientists	02 6584 471	Levin 16

Pages

Upcoming Events

Yesterday

Nonlinear Physics seminar: "Statistical mechanics of wide but finite artificial neural networks"
- Start time: 12:00pm
- End date: Wednesday, December 6th
- End time: 01:30pm
- Lecturer: Mr. Gadi Naveh (HUJI)Abstract: Over the past decade, Deep neural networks (DNNs) have brought significant advancements to various fields of science and technology. Key to their success is feature learning, enabling them to extract useful representations from raw data. However, DNNs remain challenging to interpret and lack a comprehensive analytical understanding. Recent studies revealed that DNNs in the infinite width limit are equivalent to Gaussian Processes (GPs), which are well-understoodanalytically but lack feature learning capabilities due to pre-determined features.In this thesis, we aim to bridge the gap between GPs and finite DNNs, using approaches inspired by theoretical physics. Employing statistical field theory, we treat the DNN output as a random object from a statistical ensemble. We performed extensive computer simulations of DNNs trained in both stylized and real-world settings. As for analytical methods, we first employ a perturbative approach to obtain the leading correction in 1/width to the GP limit. Subsequently, we develop a self-consistent mean-field type theory, providing a GP description of finite DNNs with a data-dependent kernel that embodies feature learning properties.
- View this event in Google Calendar
Position at Kyoto
- Start time: 02:00pm
- End date: Wednesday, December 6th
- End time: 03:00pm
- View this event in Google Calendar

Search form

Nonlinear Physics

Pages