Abstract:Neural networks are central to many emerging technologies, but verifying their correctness remains a major challenge. It is known that network outputs can be sensitive and fragile to even small input perturbations, thereby increasing the risk of unpredictable and undesirable behavior. Fast and accurate verification of neural networks is therefore critical to their widespread adoption, and in recent years, various methods have been developed as a response to this problem. In this paper, we focus on improving semidefinite programming (SDP) based techniques for neural network verification. Such techniques offer the power of expressing complex geometric constraints while retaining a convex problem formulation, but scalability remains a major issue in practice. Our starting point is the DeepSDP framework proposed by Fazlyab et al., which uses quadratic constraints to abstract the verification problem into a large-scale SDP. However, solving this SDP quickly becomes intractable when the network grows. Our key observation is that by leveraging chordal sparsity, we can decompose the primary computational bottleneck of DeepSDP -- a large linear matrix inequality (LMI) -- into an equivalent collection of smaller LMIs. We call our chordally sparse optimization program Chordal-DeepSDP and prove that its construction is identically expressive as that of DeepSDP. Moreover, we show that additional analysis of Chordal-DeepSDP allows us to further rewrite its collection of LMIs in a second level of decomposition that we call Chordal-DeepSDP-2 -- which results in another significant computational gain. Finally, we provide numerical experiments on real networks of learned cart-pole dynamics, showcasing the computational advantage of Chordal-DeepSDP and Chordal-DeepSDP-2 over DeepSDP.

Scalable Analysis of Linear Networked Systems Via Chordal Decomposition

Robust Stability Analysis of Networked Control Systems

Scalable Design of Structured Controllers Using Chordal Decomposition

A Chordal Decomposition Approach to Scalable Design of Structured Feedback Gains over Directed Graphs

A Decentralized Output Feedback Controller for a Class of Large-Scale Interconnected Nonlinear Systems

Early-stage power grid analysis using a supenode-like formulation of circuit equations

Chordal Sparsity for SDP-based Neural Network Verification

Distributed Learning of Neural Lyapunov Functions for Large-Scale Networked Dissipative Systems

Decomposition of Nonlinear Dynamical Networks via Comparison Systems

Convergence Analysis of Distributed Kalman Filtering for Relative Sensing Networks

Statistical Learning for Analysis of Networked Control Systems over Unknown Channels

Complexity of Chordal Conversion for Sparse Semidefinite Programs with Small Treewidth

A New Decomposition Strategy for Analyzing Large-Scale Systems Such As Power Systems

Stability Analysis of Distributed Estimators for Large-Scale Interconnected Systems: Time-Varying and Time-Invariant Cases

Chordal and Factor-Width Decompositions for Scalable Semidefinite and Polynomial Optimization.

Exploiting Chordal Sparsity for Fast Global Optimality with Application to Localization

A Unified Approach to Scalable Spectral Sparsification of Directed Graphs

A Framework of Stability Analysis for Multi-agent Systems on Arbitrary Topology Graph: Linear Systems

Chordal Decomposition in Operator-Splitting Methods for Sparse Semidefinite Programs

Decomposition Methods for Large-Scale Semidefinite Programs with Chordal Aggregate Sparsity and Partial Orthogonality

Robust Analysis for the Distributed Network with the Subsystems of Lure Systems