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


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


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

Computer Science > Machine Learning

Title: Polarity is all you need to learn and transfer faster

Abstract: Natural intelligences (NIs) thrive in a dynamic world - they learn quickly, sometimes with only a few samples. In contrast, artificial intelligences (AIs) typically learn with a prohibitive number of training samples and computational power. What design principle difference between NI and AI could contribute to such a discrepancy? Here, we investigate the role of weight polarity: development processes initialize NIs with advantageous polarity configurations; as NIs grow and learn, synapse magnitudes update, yet polarities are largely kept unchanged. We demonstrate with simulation and image classification tasks that if weight polarities are adequately set a priori, then networks learn with less time and data. We also explicitly illustrate situations in which a priori setting the weight polarities is disadvantageous for networks. Our work illustrates the value of weight polarities from the perspective of statistical and computational efficiency during learning.
Comments: ICML camera-ready
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE); Neurons and Cognition (q-bio.NC)
Cite as: arXiv:2303.17589 [cs.LG]
  (or arXiv:2303.17589v2 [cs.LG] for this version)

Submission history

From: Qingyang Wang [view email]
[v1] Wed, 29 Mar 2023 14:43:04 GMT (8551kb,D)
[v2] Tue, 30 May 2023 19:26:34 GMT (8579kb,D)

Link back to: arXiv, form interface, contact.