Learn More
We study the performance of an on-line algorithm for learning di-chotomies, with a dynamical error-dependent learning rate. The asymp-totic scaling form of the solution to the associated Markov equations is derived, assuming certain smoothness conditions. We show that the system converges to the optimal solution and the generalization error vanishes(More)
The performance of on line algorithms for learning dichotomies is studied In on line learn ing the number of examples P is equivalent to the learning time since each example is presented only once The learning curve or generalization error as a function of P depends on the schedule at which the learning rate is lowered For a target that is a perceptron rule(More)
The eKect of the structure of the input distribution on the complexity of learning a pattern classification task is investigated. Using statistical mechanics, we study the performance of a winnertake-all machine at learning to classify points generated by a mixture of K Gaussian distributions ("clusters" ) in R with intercluster distance u (relative to the(More)
Humans and animals often acquire knowledge about their environment and shape their responses to it through learning, The process of learning differs drastically from those used in conventional machine information processing systems, \,yhat are the powers and limitations of extracting rules and discovering regularities in data by learning from examples? Is(More)
  • 1