We propose and analyze a distribution learning algorithm for variable memory length Markov processes. These processes can be described by a subclass of probabilistic nite automata...
Abstract The cerebral cortex utilizes spatiotemporal continuity in the world to help build invariant representations. In vision, these might be representations of objects. The temp...
Simon M. Stringer, G. Perry, Edmund T. Rolls, J. H...
The goal of semi-supervised learning (SSL) methods is to reduce the amount of labeled training data required by learning from both labeled and unlabeled instances. Macskassy and Pr...
It is well known that many hard tasks considered in machine learning and data mining can be solved in a rather simple and robust way with an instanceand distance-based approach. In...
As robots become more common, it becomes increasingly useful for them to communicate and effectively share knowledge that they have learned through their individual experiences. L...