Quality of internal representation shapes learning performance in feedback neural networks
Lee Susman, Francesca Mastrogiuseppe, Naama Brenner, Omri Barak
Abstract
The authors find a resonance frequency at which feedback neural networks reproduce a sinusoidal wave, and show that network performance is determined by the robustness of the corresponding internal representation.
Topics & Concepts
Representation (politics)Computer scienceQuality (philosophy)Artificial neural networkInternal modelArtificial intelligenceMachine learningControl (management)PhysicsPolitical scienceLawPoliticsQuantum mechanicsNeural Networks and Reservoir ComputingNeural Networks and ApplicationsAdvanced Memory and Neural Computing