Demystifying Lipschitz verification: positive matrices, negative results

TL;DR

This paper explores the computational challenges of Lipschitz verification in neural networks, showing NP-hardness barriers and proposing regularization techniques to improve bounds.

Contribution

It reveals the structural reasons behind verification difficulties and introduces bias-free trigonometric layers with regularization for tighter Lipschitz bounds.

Findings

SDP-based bounds can inherit trivial bound failures

Reachability estimation is NP-hard, hindering polynomial algorithms

Bias-free trigonometric layers with regularization improve Lipschitz bounds

Abstract

The global Lipschitz constant of a neural network is related to robustness and generalization, yet unlike in many classical models, it is not plainly legible from the parameters. This has motivated sophisticated verification algorithms, especially semidefinite programming (SDP) based on incremental quadratic constraints on the activation functions, to improve on the fast but often loose product of layerwise Lipschitz constants (the trivial bound). We ask why Lipschitz verification is a problem in the first place. Our answer is that the difficulty is structural: estimating a network's Lipschitz constant requires knowing which hidden states are reachable, and reachability is NP-hard. If P!=NP, then reachability is a barrier to any polynomial-time algorithm. Through explicit constructions, we show that this blindness can force SDP-based bounds to inherit the same qualitative failures as the trivial bound, including but not limited to polynomial per-layer conservatism. We show that the difficulties of NP-hard questions are not isolated to worst-case computational reductions, but actually afflict every instance of the verification problem. Thus SDP is not sufficient for Lipschitz verification. We also argue that it is not necessary: several apparent failures of the trivial bound arise from removable parameterization pathologies, and can be mitigated by optimizing or regularizing the trivial bound itself. We demonstrate this claim via a "spherical cow" linear model and numerical proofs of concept. While the main contribution is theoretical and negative, we finally motivate a novel form of trigonometric layers that do not need biases for universal approximation. Combined with trivial bound regularization, they make the trivial bound provably and practically tight.