Preprints, Working Papers, ... Year : 2024

Training of Physical Neural Networks

Ali Momeni
  • Function : Author
Babak Rahmani
  • Function : Author
Benjamin Scellier
  • Function : Author
Logan G Wright
  • Function : Author
Peter L Mcmahon
  • Function : Author
Clara C Wanjura
  • Function : Author
Yuhang Li
  • Function : Author
Anas Skalli
  • Function : Author
Natalia G Berloff
  • Function : Author
Tatsuhiro Onodera
  • Function : Author
Ilker Oguz
  • Function : Author
Francesco Morichetti
  • Function : Author
Manuel Le Gallo
  • Function : Author
Abu Sebastian
  • Function : Author
Azalia Mirhoseini
  • Function : Author
Cheng Zhang
  • Function : Author
Danijela Marković
  • Function : Author
Daniel Brunner
  • Function : Author
Christophe Moser
  • Function : Author
Sylvain Gigan
Florian Marquardt
  • Function : Author
Aydogan Ozcan
  • Function : Author
Julie Grollier
  • Function : Author
Andrea J Liu
  • Function : Author
Demetri Psaltis
  • Function : Author
Andrea Alù
  • Function : Author
Romain Fleury
  • Function : Author
  • PersonId : 1185373

Abstract

Physical neural networks (PNNs) are a class of neural-like networks that leverage the properties of physical systems to perform computation. While PNNs are so far a niche research area with small-scale laboratory demonstrations, they are arguably one of the most underappreciated important opportunities in modern AI. Could we train AI models 1000x larger than current ones? Could we do this and also have them perform inference locally and privately on edge devices, such as smartphones or sensors? Research over the past few years has shown that the answer to all these questions is likely "yes, with enough research": PNNs could one day radically change what is possible and practical for AI systems. To do this will however require rethinking both how AI models work, and how they are trained - primarily by considering the problems through the constraints of the underlying hardware physics. To train PNNs at large scale, many methods including backpropagation-based and backpropagation-free approaches are now being explored. These methods have various trade-offs, and so far no method has been shown to scale to the same scale and performance as the backpropagation algorithm widely used in deep learning today. However, this is rapidly changing, and a diverse ecosystem of training techniques provides clues for how PNNs may one day be utilized to create both more efficient realizations of current-scale AI models, and to enable unprecedented-scale models.

Fichier principal
Vignette du fichier
2406.03372v1 (1).pdf (2.65 Mo) Télécharger le fichier
Origin Files produced by the author(s)
Licence

Dates and versions

hal-04775753 , version 1 (10-11-2024)

Licence

Identifiers

Cite

Ali Momeni, Babak Rahmani, Benjamin Scellier, Logan G Wright, Peter L Mcmahon, et al.. Training of Physical Neural Networks. 2024. ⟨hal-04775753⟩
64 View
438 Download

Altmetric

Share

  • More