Robert Holte

Professor Emeritus, Department of Computing Science
University of Alberta

Professor Holte’s research aims to develop scientific understanding and improved algorithms in the field of Artificial Intelligence, with a focus in the areas of Machine Learning and Heuristic Search.

A machine learning algorithm uses training data, or past experience, to improve its performance at a specific task. His two most important contributions to machine learning are the 1R (OneR) system, and Cost Curves. 1R is an extremely simple system, making classifications based on just a single feature. Often, it is surprisingly accurate. It has been cited more than 2500 times by research work from many different fields of study, either as a classifier or a feature selection method. Cost Curves are used to evaluate and compare binary classifiers when misclassification costs are unequal or one of the classes is much rarer than the other. It is superior to any other method and, like 1R, has become a useful tool in a variety of fields of study.

A heuristic search algorithm is a fast method for finding optimal paths between two nodes in a graph. For example, if the graph is a road map, a heuristic search algorithm could be used to quickly find the shortest route from your current location to a particular destination. Heuristic search is faster than standard methods because it uses an estimate of the distance to the goal, called a heuristic function. His two main contributions in this field are, (a) devising ways to create heuristic functions automatically, and (b) bidirectional search. He has explored many approaches to do (a): machine learning and abstraction – which means mapping the search space to an abstract search space and then using true distances in the abstract space as estimates of the distances in the original search space. Regarding (b), he has made substantial theoretical analysis to advance our understanding of bidirectional search, and has developed the MM algorithm, whose forward and backward search searches are guaranteed to “meet in the middle”, a property that had eluded researchers for 50 years.

In addition to these impressive research results, he also served for many years as the Editor-in-Chief of “Machine Learning”, which was then unquestionably the top journal of the field. He was also one of the founders of the Alberta Machine Intelligence Institute, which is one of the world’s top centres for Artificial Intelligence.

 

EN