Structure Is Not Enough: Leveraging Behavior for Neural Network Weight Reconstruction

Léo Meynent, Ivan Melev, Konstantin Schürholt, Goeran Kauermann, Damian Borth

Track: long paper (up to 8 pages)

Keywords: probing-based loss function, self-supervised learning, autoencoders

TL;DR: We show that training weight-space autoencoders with a behavioral loss component improves neural network weights reconstruction and generation

Submission Number: 35