|
Abstract:
|
Over the last decade, newly developed machine learning methods like Boosting and Support Vector Machine (SVM) have focused on univariate classification and regression. Can these results be transferred to multivariate prediction problems, where the goal is to predict complex objects like trees, sequences, or orderings? Such problems arise, for example, when a natural language parser needs to predict the parse tree for a sentence, when a navigation assistant needs to predict a route, or when a search engine needs to predict a ranking. This talk will discuss an SVM approach to predicting complex objects. It generalizes the idea of margins to complex prediction problems and a large range of loss functions. While the resulting training problems have exponential size, there is a simple algorithm that allows training in polynomial time. Empirical results will be given for several examples.
|
