A Biological Gradient Descent for Prediction Through a Combination of STDP and Homeostatic Plasticity

• DOI: 10.1162/neco_a_00512
• Publication Date: 2013-09-03T15:41:03Z
• Authors: Mathieu N. Galtier; Gilles Wainrib
• Keywords: Homeostatic plasticity; Stochastic Gradient Descent

Identifying, formalizing, and combining biological mechanisms that implement known brain functions, such as prediction, is a main aspect of research in theoretical neuroscience. In this letter, the spike-timing-dependent plasticity homeostatic plasticity, combined an original mathematical formalism, are shown to shape recurrent neural networks into predictors. Following rigorous treatment, we prove they online gradient descent distance between network activity its stimuli. The convergence equilibrium, where can spontaneously reproduce or predict stimuli, does not suffer from bifurcation issues usually encountered learning networks.