Here we grow again! Meet Introspective Systems’ Senior Research Scientist, Clare Bates Congdon, who came to us via Bowdoin College.
Clare has been working in Machine Learning for over 25 years and teaching Computer Science for 20. Her expertise includes machine learning, artificial intelligence, genetic algorithms, and complex adaptive systems. Her work in machine learning and bioinformatics has been funded by both the National Science Foundation and the National Institutes of Health. Her work in intelligent agents and games has won four international competitions. She is active in international professional organizations and has been Chair of the IEEE CIS Technical Committee on Bioinformatics and Bioengineering, and Vice Chair of the IEEE CIS Technical Committee on Games. Congdon holds a BA in Mathematics from Wesleyan University and an MS and Ph.D. in Computer Science and Engineering from the University of Michigan.
You recently left a teaching position at Bowdoin College, what led you to make the move from academia to the private sector?
I’d been a college professor for 20 years, doing extensive research in artificial intelligence and machine learning as applied in particular to bioinformatics and to agents work (including games). It’s a challenge to juggle the competing demands of being a good teacher with being a good researcher, so I decided to double down on the research. Having the opportunity to work at Introspective Systems made this decision even easier. I have always enjoyed collaborating with others, and the IS environment is the perfect balance of aligning my previous experience and working in a team-centered environment while providing new intellectual challenges. Additionally, there is a great appeal for me to live and work in Portland.
You recently completed work on a large project with researchers at Dartmouth Medical School. Can you describe this collaboration?
This project was funded by the National Institutes of Health and focused on the interdisciplinary work of Quantitative Biology. The project spanned sites in the northeastern states and involved bench scientists, statisticians, and computer scientists, and was particularly notable for the interdisciplinary nature of the projects and collaborations. My team, comprised of myself and student researchers, developed a system for computational identification of candidate regulatory regions in noncoding DNA. This is the part of the DNA that used to be called “junk”, but we now know that within these long stretches of DNA are important elements that affect how genes work. The tools developed in my 5-year, 1.2-million-dollar project, expedite the expensive and time-consuming process of identifying these regions in the biological laboratory.
You won four international prizes for your work with intelligent agents and games, wow! Can you describe the work you did to earn that?
When I was at the University of Southern Maine (USM), I had just taught a course in Artificial Intelligence when I heard about a competition to create an artificial player (an agent) for Ms. PacMan. In the first year of the competition, no one “won”, meaning that the entrant’s agents could not outperform the initial code provided by the organizers. Student Alan Fitzgerald (now at CashStar) was lead on our small team; we developed some new approaches for looking at the game and implementing an artificial player. We competed in the second year of this competition, held in Hong Kong, and won!
Unsurprisingly, there was great interest in working with AI and Games at USM, and I had several other students participate in projects to play Ms. PacMan, Unreal, and Mario. Often, these students were not interested in actually competing, but yes, we won first-place honors at international competitions four times with this line of research.
A lot of people, especially non-technical folks, think of games as purely for entertainment. What are some of the “scientific” contributions from this field of study?
Prior to USM, I had done some work with AI and robotics (and similarly was part of a team that won a national competition in grad school). In some ways, working with computer games resembles working on robotics in simulation: There is an agent navigating an environment, and the agent is autonomous, meaning that it makes its own decisions about what to do next. In many cases, the environment of a game is less complex than the real world, but there are some games (such as StarCraft) where the complexity of the environment and necessary decision making is on par with “real world” scenarios. In other words, games can provide a robust environment for research in intelligent agents.
What are some of the projects at Introspective Systems you’d like to investigate?
I have a deep background in intelligent agents and am looking forward to working on the application of xGraph to StarCraft. I have also been delighted over the course of my career to work on science problems and to help understand biological, genetic, and medical data, so I am looking forward to more projects along those lines. In general, I have extensive experience with developing machine learning approaches to better understand data and to implement adaptive agents, and I’m looking forward to contributing this skill set to the existing talent at Introspective Systems.