English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

We show that discrete synaptic weights can be efficiently used for learningin large scale neural systems, and lead to unanticipated computationalperformance. We focus on the representative case of learning random patternswith binary synapses in single layer networks. The standard statisticalanalysis shows that this problem is exponentially dominated by isolatedsolutions that are extremely hard to find algorithmically. Here, we introduce anovel method that allows us to find analytical evidence for the existence ofsubdominant and extremely dense regions of solutions. Numerical experimentsconfirm these findings. We also show that the dense regions are surprisinglyaccessible by simple learning protocols, and that these synaptic configurationsare robust to perturbations and generalize better than typical solutions. Theseoutcomes extend to synapses with multiple states and to deeper neuralarchitectures. The large deviation measure also suggests how to design novelalgorithmic schemes for optimization based on local entropy maximization.

Subdominant Dense Clusters Allow for Simple Learning and High Computational Performance in Neural Networks with Discrete Synapses