LGCRCVApr 25, 2023

Lyapunov-Stable Deep Equilibrium Models

arXiv:2304.12707v311 citationsh-index: 22
Originality Incremental advance
AI Analysis

This addresses stability and robustness problems for users of deep equilibrium models in adversarial settings, representing an incremental advance.

The paper tackles the stability issue of deep equilibrium models by proposing LyaDEQ, a model with provable Lyapunov stability to resist minor perturbations, and demonstrates significant robustness improvements under adversarial attacks.

Deep equilibrium (DEQ) models have emerged as a promising class of implicit layer models, which abandon traditional depth by solving for the fixed points of a single nonlinear layer. Despite their success, the stability of the fixed points for these models remains poorly understood. By considering DEQ models as nonlinear dynamic systems, we propose a robust DEQ model named LyaDEQ with guaranteed provable stability via Lyapunov theory. The crux of our method is ensuring the Lyapunov stability of the DEQ model's fixed points, which enables the proposed model to resist minor initial perturbations. To avoid poor adversarial defense due to Lyapunov-stable fixed points being located near each other, we orthogonalize the layers after the Lyapunov stability module to separate different fixed points. We evaluate LyaDEQ models under well-known adversarial attacks, and experimental results demonstrate significant improvement in robustness. Furthermore, we show that the LyaDEQ model can be combined with other defense methods, such as adversarial training, to achieve even better adversarial robustness.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes