Source author record

Gaurang Mahajan

Gaurang Mahajan appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

nlin.AO cond-mat.dis-nn cond-mat.stat-mech Neurons and Cognition

Catalog footprint

What is connected

3works

4topics

2close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2011arXiv

Competing synapses with two timescales: a basis for learning and forgetting

Competitive dynamics are thought to occur in many processes of learning involving synaptic plasticity. Here we show, in a game theory-inspired model of synaptic interactions, that the competition between synapses in their weak and strong states gives rise to a natural framework of learning, with the prediction of memory inherent in a timescale for `forgetting' a learned signal. Among our main results is the prediction that memory is optimized if the weak synapses are really weak, and the strong synapses are really strong. Our work admits of many extensions and possible experiments to test its validity, and in particular might complement an existing model of reaching, which has strong experimental support.

preprint2011arXiv

Learning with a network of competing synapses

Competition between synapses arises in some forms of correlation-based plasticity. Here we propose a game theory-inspired model of synaptic interactions whose dynamics is driven by competition between synapses in their weak and strong states, which are characterized by different timescales. The learning of inputs and memory are meaningfully definable in an effective description of networked synaptic populations. We study, numerically and analytically, the dynamic responses of the effective system to various signal types, particularly with reference to an existing empirical motor adaptation model. The dependence of the system-level behavior on the synaptic parameters, and the signal strength, is brought out in a clear manner, thus illuminating issues such as those of optimal performance, and the functional role of multiple timescales.

preprint2010arXiv

Competing with oneself: Introducing self-interaction in a model of competitive learning

A competitive learning model was introduced in Ref. 1 (A. Mehta and J. M. Luck, Phys. Rev. E 60, 5, 1999), in which the learning is outcome-related. Every individual chooses between a pair of existing strategies or types, guided by a combination of two factors: tendency to conform to the local majority, \em{and} a preference for the type with higher perceived success \em{among its neighbors}, based on their relative outcomes. Here, an extension of the \em{interfacial model} of Ref. 1 is proposed, in which individuals additionally take into account their \em{own} outcomes in arriving at their outcome-based choices. Three possible update rules for handling bulk sites are considered. The corresponding phase diagrams, obtained at coexistence, show systematic departures from the original interfacial model. Possible relationships of these variants with the \em{cooperative model} of Ref. 1 are also touched upon.