Student & Faculty Research

Students and faculty in the computer science department are very active in research. There are numerous student-faculty research projects, independent projects, and group projects. Students present their work at different research forums, both on-campus and off-campus, and there are several faculty research projects with active student participation.

Professor Christopher Andrews works in the areas of visual analytics and computational art. Recent work has focused on supporting the development of bespoke analytic tools. This culminated in the development of MiddGuard, a flexible web framework that serves as scaffolding for the development of highly interactive collaborative visual analytics tools. Primarily developed by Dana Silver '17, the framework has been used to compete in several analytic contests at the VAST conference, winning an award for "Integrated Analytic Environment". He has also been working with students on bridging the gap between computer science and the arts through explorations of generative and evolutionary artwork.

Professor Amy Briggs works on expanding access to computer science education in K-12. She is one of the creators of the AP Computer Science Principles Course as co-PI on the National Science Foundation award to the College Board on broadening participation in computing. She has worked collaboratively with students in organizing outreach efforts with middle school and high school girls. She supported the creation of the WiCS++ group and continues to serve as one of the faculty advisers.

Professor Ananya Christman's current research involves developing algorithms for optimization problems. One problem is Online-Dial-a-Ride where a server travels to serve customer requests for rides within a given time limit. Each request has a revenue and the goal is to maximize total revenue.  Another problem is to find an assignment of applications to servers on a distributed system such that the system is robust. The goal is to find such an assignment that  uses the minimum number of servers.  A recent publication co-authored with 3 Middlebury undergraduates is the paper "Revenue Maximization in Online Dial-A-Ride," which appeared in the Workshop on Algorithmic Approaches for Transportation Modelling, Optimization, and Systems (ATMOS) in September 2017.

In recent years, Professor Matthew Dickerson has turned his attention from traditional problems in computational geometry and geo-spatial computing to the growing field of agent-based modeling, also known as multi-agent simulation—a field closely related to what biologists refer to as individual-based ecology and social scientists refer to as complex adaptive systems. He has worked with students on computer modeling/simulation projects related to forest succession and competition of big-leaf mahogany in the Amazon, population dynamics of transient killer whales in southeast Alaska, and the impact of invasive trout on stream ecology and forest-stream dynamics. His interest in computer modeling in ecology has also corresponded with work outside of computer science as a narrative non-fiction nature and environmental writer.

Professor Jason Grant is researching topics in the area of biometric recognition applications and computer vision. His previous work in biometrics explored the proximity of identical twins through facial recognition and classification of extrinsic characteristics revealed in face images, such as gender, ethnicity, and expression, based upon the relationship (proximity scores) between face images. Future work in this area aim to use proximity data to prune search spaces and exploit algorithm biases towards certain extrinsic features. Additionally, he is actively working on methods for real-time detection of abnormal crowd behavior. While current methods are restricted to fixed camera video, he is pursuing approaches for abnormal crowd behavior detection using drones.

Professor Peter Johnson works in the area of systems and network security.  Under his guidance, Kit Tse '16.5 produced a framework for verifying protocol sessions, including a proof of concept for TCP, the protocol that undergirds the entire Internet.  She presented this work at the Fourth Workshop on Language-Theoretic Security in May 2017.

Professor Shelby Kimmel designs quantum algorithms for graph connectivity. Studying this problem is exciting because it is both useful for many real world applications, but also it is giving insight into new ways to create and analyze quantum algorithms. She also aims to help the experimentalists building quantum computers by coming up with better ways to characterize errors in their systems. To accomplish this, she uses tools from signal processing, like compressed sensing, and classic quantum techniques, like phase estimation.

Professor Michael Linderman’s research interests include heterogeneous computing (GPGPU), genome analysis and genomics education for the public, patients and providers. He is the PI of a NIH-funded project to create a genomics knowledge measure and is pursuing several genome analysis projects with Middlebury students. Current projects include: DECA, a software package for CNV calling that began as a senior thesis project; MySeq, a web application for exploring your own whole genome sequencing data; and a novel method for identifying disease causing mutations using simulation.

Professor Daniel Scharstein is known in the computer vision community for the "Middlebury Benchmarks," a collection of datasets and online leaderboards that were developed in collaboration with student research assistants over the last 15 years.  Researchers around the world use these benchmarks to test their stereo vision and 3D reconstruction methods.  Professor Scharstein also works on novel algorithms for stereo vision. A recent collaboration with Dylan Quenneville '18 resulted in a paper "Mondrian Stereo," which Dylan presented at the International Conference for Image Processing (ICIP) in China in September 2017.