Kai-Wei Chang, Gourab Kundu, Vivek Srikumar and Dan Roth
AAAI Tutorial Forum 2016.


Many prediction problems require assigning values to multiple interdependent variables. The relationships between these variables could represent a sequence, a set of clusters, or in the general case, a graph. Over past decades, structured prediction models for such problems have demonstrated success in a range of applications, including natural language processing, computer vision and computational biology. However, the high computational cost often limits both the expressive power of the models and the size of the data that can be handled. Therefore, designing efficient inference and learning algorithms for these models is a key challenge.

In this tutorial, beyond introducing the algorithmic approaches, we will discuss ideas that result in significant improvements both in the learning and the inference stages of structured prediction models. In particular, we will discuss the use of caching techniques to reuse computations and methods for decomposing complex structures, along with learning procedures that use these approaches to simplify the learning stage. We will also present a formulation that captures similarities between structured labels using distributed representations. Participants will learn about the recent trends in this domain, tools developed in this area, and how they can be applied to AI applications.


Bib Entry

  title = {{Learning and Inference in Structured Prediction Models}},
  author = { Kai-Wei Chang and Gourab Kundu and Dan Roth and Vivek Srikumar},
  booktitle = {AAAI-16 Tutorial Forum},
  month = {February},
  year = {2016}