Noise sensitivity and stability of deep neural networks for binary classification

DOI10.1016/J.SPA.2023.08.003MaRDI QIDQ6080378zbMATH OpenOpenAlexFDO

Authors Johan Jonasson, Jeffrey E. Steif, Olof Zetterqvist

Publication date 30 October 2023

Published in Stochastic Processes and their Applications (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/2308.09374

zbMATH Keywords

Boolean functionsnoise sensitivitynoise stabilityfeed-forward neural networksdeep neural networks

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Artificial neural networks and deep learning (68T07) Interacting random processes; statistical mechanics type models; percolation theory (60K35) Combinatorial probability (60C05) Boolean functions (06E30)

Recommendations

• Stability for the training of deep neural networks and other classifiers
• A noise-based stabilizer for convolutional neural networks
• Stability of analogue neural classification networks
• Stability of Neural Networks for Slightly Perturbed Training Data Sets
• Adversarial noise attacks of deep learning architectures: stability analysis via sparse-modeled signals
• Noise-enhanced convolutional neural networks
• Stability of Deep Neural Networks via Discrete Rough Paths

Cites work

• Analysis of Boolean Functions
• Approximation by superpositions of a sigmoidal function
• Noise sensitivity of Boolean functions and applications to percolation
• Noise stability of weighted majority
• On the rate of convergence in quenched Voronoi percolation
• Partially observed Boolean sequences and noise sensitivity
• Quenched Voronoi percolation

Cited in

(1)

• Fragility, robustness and antifragility in deep learning

This page was built for publication: Noise sensitivity and stability of deep neural networks for binary classification

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6080378)