{ "id": "1211.3024", "version": "v1", "published": "2012-11-13T15:44:13.000Z", "updated": "2012-11-13T15:44:13.000Z", "title": "Generalization learning in a perceptron with binary synapses", "authors": [ "Carlo Baldassi" ], "comment": "16 pages, 4 figures", "journal": "Journal of Statistical Physics 136 (2009) 902-916", "doi": "10.1007/s10955-009-9822-1", "categories": [ "cond-mat.dis-nn" ], "abstract": "We consider the generalization problem for a perceptron with binary synapses, implementing the Stochastic Belief-Propagation-Inspired (SBPI) learning algorithm which we proposed earlier, and perform a mean-field calculation to obtain a differential equation which describes the behaviour of the device in the limit of a large number of synapses N. We show that the solving time of SBPI is of order N*sqrt(log(N)), while the similar, well-known clipped perceptron (CP) algorithm does not converge to a solution at all in the time frame we considered. The analysis gives some insight into the ongoing process and shows that, in this context, the SBPI algorithm is equivalent to a new, simpler algorithm, which only differs from the CP algorithm by the addition of a stochastic, unsupervised meta-plastic reinforcement process, whose rate of application must be less than sqrt(2/(\\pi * N)) for the learning to be achieved effectively. The analytical results are confirmed by simulations.", "revisions": [ { "version": "v1", "updated": "2012-11-13T15:44:13.000Z" } ], "analyses": { "subjects": [ "F.2.2", "I.2.6", "I.5.1" ], "keywords": [ "binary synapses", "generalization learning", "unsupervised meta-plastic reinforcement process", "generalization problem", "stochastic" ], "tags": [ "journal article" ], "publication": { "journal": "Journal of Statistical Physics", "year": 2009, "month": "Sep", "volume": 136, "number": 5, "pages": 902 }, "note": { "typesetting": "TeX", "pages": 16, "language": "en", "license": "arXiv", "status": "editable", "adsabs": "2009JSP...136..902B" } } }