Tal Grossman

Learn More
Recently, there has been considerable interest in deriving and applying knowledge-based, empirical potential functions for proteins. These empirical potentials have been derived from the statistics of interacting, spatially neighboring residues, as may be obtained from databases of known protein crystal structures. In this paper we employ neural networks to(More)
A neural network model, the INN (Inverted Neurons Network), is applied to the Maximum Clique problem. First, I describe the INN model and how it implements a given graph instance. The model has a threshold parameter t, which determines the character of the network stable states. As shown in an earlier work 5], the stable states of the network correspond to(More)
A new learning algorithm for feedforward networks, learning by choice of intern al represent at ions (C HIR), was recently introduced [1,2]. W hereas many algor it hm s red uce th e learning proce ss to minimizing a cost function over t he weights, our method treats th e internal representations as the funda ment al ent it ies to be determi ned. T he algo(More)
A new learning algorithm, learning by choice of int ern al repr esentations (CHIR), was recently int roduced. Th e basic version of thi s algoriths was developed for a two-layer, single-out put , feedforward network of binary neurons . This paper presents a generalized version of the CHIR algorithm th at is capable of t raining mult ipleoutput net works. A(More)
We study the extent to which xing the second layer weights reduces the capacity and generalization ability of a two-layer perceptron. Architectures with N inputs, K hidden units and a single output are considered, with both overlapping and non-overlapping receptive elds. We obtain from simulations one measure of the strength of a network-its critical(More)
In this paper we analyze the average behavior of the Bayes-optimal and Gibbs learning algorithms. We do this both for oo-training-set error and conventional IID error (for which test sets overlap with training sets). For the IID case we provide a major extension to one of the better known results of 7]. We also show that expected IID test set error is a(More)