Deconstructing Data Reconstruction: Multiclass, Weight Decay and General Losses

Gon Buzaglo, Niv Haim, Gilad Yehudai, Gal Vardi, Yakir Oz, Yaniv Nikankin, michal Irani

Keywords: memorization, data reconstruction, implicit bias

TL;DR: We extend previous reconstruction methods from trained neural networks, including reconstruction from multiclass and convolutional networks, and analyze the various factors which enable reconstructability

Abstract:

Memorization of training data is an active research area, yet our understanding of the inner workings of neural networks is still in its infancy.
Recently, Haim et al. 2022 proposed a scheme to reconstruct training samples from multilayer perceptron binary classifiers, effectively demonstrating that a large portion of training samples are encoded in the parameters of such networks.
In this work, we extend their findings in several directions, including reconstruction from multiclass and convolutional neural networks.
We derive a more general reconstruction scheme which is applicable to a wider range of loss functions such as regression losses.
Moreover, we study the various factors that contribute to networks' susceptibility to such reconstruction schemes.
Intriguingly, we observe that using weight decay during training increases reconstructability both in terms of quantity and quality.
Additionally, we examine the influence of the number of neurons relative to the number of training samples on the reconstructability.
Code: https://github.com/gonbuzaglo/decoreco

Submission Number: 2783