I am a member of the Knowledge Discovery Laboratory, advised by David Jensen. I am broadly interested in using causal reasoning to develop robust systems for AI-driven decision-making. My research studies the identification of anomalous or malicious activity, development of methods for resilience and few-shot transfer of decision-making policies, and construction of empirical designs for evaluating behavior of complex (software) systems. My recent work is focused in the sequential decision-making setting and studies the problem of how machine-learning and AI systems can respond intelligently to unexpected inputs, anomalies, and other forms of novelty in the environment. I am particularly motivated by applications in fraud prevention, cybersecurity, and climate risk assessment.
Feel free to check out my projects, resume, or send me mail at .
Recent Highlights:
Other News:
- Our paper Exploratory Not Explanatory: Counterfactual Analysis of Saliency Maps for Deep Reinforcement Learning (coauthors Akanksha Atrey and David Jensen) was accepted at ICLR 2020. (Dec 2019)
- Our paper Let's Play Again: Variability of Deep Reinforcement Learning Agents in Atari Environments (coauthors Emma Tosch, John Foley, and David Jensen) was accepted to the NeurIPS Critiquing and Correcting Trends workshop and featured in a spotlight talk. (Dec 2018)
- Our work ToyBox: Better Atari Environments for Testing Reinforcement Learning Agents (coauthors John Foley, Emma Tosch, and David Jensen) was accepted to the NeurIPS Systems for ML workshop. (Dec 2018)
- I was accepted as a 2018 Data Science for Social Good Fellow. For 3 months, I worked with a small team of fellows in partnership with AllianceChicago, a national network of 44 community health centers serving the least resourced members of their communities, on a project to support early diabetes screenings by identifying patients at risk of developing the disease in the next 3 years. Our code will be made publicly available. You can read an excerpt of my application in the UMass announcement. (August 2018)
- Completed a Masters degree and accepted into candidacy. My Masters thesis on A/B Testing in Networks with Adversarial Nodes received a 2017 CICS Outstanding Synthesis Project Award. (February 2018)
- Our work Evaluating Saliency Maps Using Interventions (coauthors Akanksha Atrey and David Jensen) was accepted as a poster at the 2019 WiML Workshop colocated with NeurIPS. (Oct 2019)
- My work with David Jensen on A/B Testing in Networks with Adversarial Members (paper / code) was accepted at the 2017 Workshop on Mining and Learning with Graphs at KDD 2017. You can check out the slides from my workshop talk or watch the video. (August 2017)
- Interned with the Human and Language Technologies group at MIT Lincoln Labs. (June 2017)
- Presented a poster, "Does Topology Matter? The Impact of Network Structure on Graph Cluster Randomization", at the 2016 Data Science Mixer hosted by the UMass Amherst Center for Data Science. I also gave a lightning talk. (October 2016)
- Participated in the 2015 Hack the Dinos challenge at the American Museum of Natural History, and featured in a New York Times article. Check out our project on GitHub! (November 2015)
- Mentored young girls age 12-15 in introductory programming using the Scratch language through the Girls, Inc. of Holyoke Eureka! program. (July 2015)
- Presented a poster on "Alternative approaches to discovering causality with additive noise models" at New England Machine Learning Day 2015. (May 2015)
|
