We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Neural Networks with Complex-Valued Weights Have No Spurious Local Minima

Authors: Xingtu Liu
Abstract: We study the benefits of complex-valued weights for neural networks. We prove that shallow complex neural networks with quadratic activations have no spurious local minima. In contrast, shallow real neural networks with quadratic activations have infinitely many spurious local minima under the same conditions. In addition, we provide specific examples to demonstrate that complex-valued weights turn poor local minima into saddle points. The activation function CReLU is also discussed to illustrate the superiority of analytic activations in complex-valued neural networks.
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2103.07287 [cs.LG]
  (or arXiv:2103.07287v1 [cs.LG] for this version)

Submission history

From: Xingtu Liu [view email]
[v1] Sun, 31 Jan 2021 10:44:38 GMT (6671kb,D)

Link back to: arXiv, form interface, contact.