On Complex Supervised Learning Problems, and On Ranking and Choice Models

While simple supervised learning problems like binary classification and regression are fairly well understood, increasingly, many applications involve more complex learning problems: more complex label and prediction spaces, more complex loss structures, or both. The first part of the talk will discuss recent advances in our understanding of such problems, including the notion of convex calibration dimension of a loss function, unified approaches for designing convex calibrated surrogates for arbitrary losses, and connections between supervised learning and property elicitation. The second part of the talk will focus on ranking and choice models. Specifically, I will describe some of our recent work on bringing together topic modeling tools from machine learning and statistics and choice modeling tools from marketing and econometrics to develop methods for automatically discovering topics or groups of similar items from choice data. I will describe results of applying the methods to real survey data on choices among vacation destinations, and among undergraduate concentrations at Harvard.