# Efficiently testing local optimality and escaping saddles for ReLU networks

arXiv: https://arxiv.org/abs/1809.10858 · HTML: https://mlsanity.com/paper/1809.10858/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Sep 2018.

## Built on by (most cited first, 4 of 4)

- [Loss Landscape of Shallow ReLU-like Neural Networks: Stationary Points, Saddle Escape, and Network Embedding](https://mlsanity.com/paper/2402.05626/) (2024)
- [Testing Stationarity Concepts for ReLU Networks: Hardness, Regularity, and Robust Algorithms](https://mlsanity.com/paper/2302.12261/) (2023)
- [On subdifferential chain rule of matrix factorization and beyond](https://mlsanity.com/paper/2410.05022/) (2024)
- [Parameterized Complexity of Stationarity Testing for Piecewise-Affine Functions and Shallow CNN Losses](https://mlsanity.com/paper/2605.10219/) (2026)

## Abstract (arXiv)

We provide a theoretical algorithm for checking local optimality and escaping saddles at nondifferentiable points of empirical risks of two-layer ReLU networks. Our algorithm receives any parameter value and returns: local minimum, second-order stationary point, or a strict descent direction. The presence of M data points on the nondifferentiability of the ReLU divides the parameter space into at most 2M regions, which makes analysis difficult. By exploiting polyhedral geometry, we reduce the total computation down to one convex quadratic program (QP) for each hidden node, O(M) (in)equality tests, and one (or a few) nonconvex QP. For the last QP, we show that our specific problem can be solved efficiently, in spite of nonconvexity. In the benign case, we solve one equality constrained QP, and we prove that projected gradient descent solves it exponentially fast. In the bad case, we have to solve a few more inequality constrained QPs, but we prove that the time complexity is exponential only in the number of inequality constraints. Our experiments show that either benign case or bad case with very few inequality constraints occurs, implying that our algorithm is efficient in most cases.