English

Matching the Statistical Query Lower Bound for $k$-Sparse Parity Problems with Sign Stochastic Gradient Descent

Machine Learning 2024-12-09 v2 Optimization and Control Machine Learning

Abstract

The kk-sparse parity problem is a classical problem in computational complexity and algorithmic theory, serving as a key benchmark for understanding computational classes. In this paper, we solve the kk-sparse parity problem with sign stochastic gradient descent, a variant of stochastic gradient descent (SGD) on two-layer fully-connected neural networks. We demonstrate that this approach can efficiently solve the kk-sparse parity problem on a dd-dimensional hypercube (kO(d)k\leq O(\sqrt{d})) with a sample complexity of O~(dk1)\tilde{O}(d^{k-1}) using 2Θ(k)2^{\Theta(k)} neurons, matching the established Ω(dk)\Omega(d^{k}) lower bounds of Statistical Query (SQ) models. Our theoretical analysis begins by constructing a good neural network capable of correctly solving the kk-parity problem. We then demonstrate how a trained neural network with sign SGD can effectively approximate this good network, solving the kk-parity problem with small statistical errors. To the best of our knowledge, this is the first result that matches the SQ lower bound for solving kk-sparse parity problem using gradient-based methods.

Keywords

Cite

@article{arxiv.2404.12376,
  title  = {Matching the Statistical Query Lower Bound for $k$-Sparse Parity Problems with Sign Stochastic Gradient Descent},
  author = {Yiwen Kou and Zixiang Chen and Quanquan Gu and Sham M. Kakade},
  journal= {arXiv preprint arXiv:2404.12376},
  year   = {2024}
}

Comments

37 pages, 7 figures, 3 tables. In NeurIPS 2024

R2 v1 2026-06-28T15:59:02.207Z