This Year's Projects

Overview

Our REU projects fall under a very wide “systems“ umbrella. In our vision of systems, practice dominates over theory — we emphasize hands-on work from the first day and—following a systems-building standpoint—balance theory and practice in order to produce tangible results for real-world problems.

Discovering the Limits of Machine Learning

What powers machine learning? The AMISTAD Lab opens the black-box to understand how machine learning works as a form of search, governed by information theoretic and statistical constraints. We prove formal results in learning and search. We will explore how search provides a unifying concept for machine learning, how information resources and dependence structures can be leveraged to move beyond memorization to true generalization, and we will probe the formal limits of learning processes. More information about our lab can be found here, as well as a brief news item about some of our recent publications.

Intelligent Music Software (Impro-Visor)

If you have an interests in both computer science and music (particularly Jazz), our intelligent music software project may appeal to you. Our Impro-Visor software includes various learning models, including grammar learning and deep learning. It also includes interaction capabilities, such as having human player trade melodies with the computer in real-time. The focus of this REU will attempt to combine interaction and learning, so that Impro-Visor becomes more of a musical companion, listening to the user and learning from him or her. Some possible variations on this theme are the user as instructor with the program as student, or the program as instructor and the user as student.

Human-robot Teaming

The mission of the HEATlab is to create new techniques for human-robot teaming--the flexible navigation and coordination of complex, inter-related activities in shared spaces. We focus on using ideas from AI to automate the scheduling and coordination of human-robot teams. We are particularly motivated by the challenge of coordinating the activities of human-robot teams in environments that require explicit cooperation to be successful. Our goal is to create human-robot teams exploit the relative strengths of humans and agents to accomplish what neither can achieve alone.

Measuring Code Complexity

Coming soon!