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 introduce a novel, biologically plausible local learning rule thatprovably increases the robustness of neural dynamics to noise in nonlinearrecurrent neural networks with homogeneous nonlinearities. Our learning ruleachieves higher noise robustness without sacrificing performance on the taskand without requiring any knowledge of the particular task. The plasticitydynamics -- an integrable dynamical system operating on the weights of thenetwork -- maintains a multiplicity of conserved quantities, most notably thenetwork's entire temporal map of input to output trajectories. The outcome ofour learning rule is a synaptic balancing between the incoming and outgoingsynapses of every neuron. This synaptic balancing rule is consistent with manyknown aspects of experimentally observed heterosynaptic plasticity, andmoreover makes new experimentally testable predictions relating plasticity atthe incoming and outgoing synapses of individual neurons. Overall, this workprovides a novel, practical local learning rule that exactly preserves overallnetwork function and, in doing so, provides new conceptual bridges between thedisparate worlds of the neurobiology of heterosynaptic plasticity, theengineering of regularized noise-robust networks, and the mathematics ofintegrable Lax dynamical systems.

plos computational biology/plos computational biology

Synaptic balancing: a biologically plausible local learning rule that  provably increases neural network noise robustness without sacrificing task  performance