Reproduce “Neural Networks with Complex-Valued Weights Have No Spurious Local Minima” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Xingtu Liu

Authors show that for one-hidden-layer networks with quadratic activations and complex weights, all local minima are global when hidden width is at least input dimension and output weights are nonzero. Width need not exceed sample count. A three-example case contrasts this with a poor real-valued local minimum; the theorem does not guarantee gradient descent finds one.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.