Will Landecker

Research

I am currently doing research as a member of Melanie Mitchell's research group. My research focuses on understanding the decisions of classifiers, particularly as they apply to computer vision systems. Rather than treating a classifier as "black box", I examine the interactions between data and classifiers in order to (1) verify that classifiers are gleaning relevant information from the data rather than spurious statistics, and (2) incorporate feedback from the classifier back into the model in order to search very large amounts of data in an adaptive, scalable way. This work combines computer vision, theoretical machine learning and data visualization.

My research interests include:

• Music Information Retrieval
• Computer Vision
• Neural Computation
• Machine Learning

Previous Work

• Convergence of the Baum Welch Algorithm and Applications to Music, 2006. Thesis, Reed College, B.A. Mathematics. Advisor: Jamie Pommersheim.
  (To see an embarrassing interview about my thesis, with my statements curtailed and arranged into a series of naïve and grandiose claims, click here. Note that the statements would still have been naïve and gradiose even without the curtailing and arranging.)

Past Courses

Spring 2009: CS 251, Discrete Structures II

Winter 2009: CS 251, Discrete Structures II

Fall 2008: CS 250, Discrete Structures

Other Interests

• Gardening, Cooking, Traveling, Rugby (union), Touch Rugby, Languages, Music, Mushroom hunting