Tom Hartvigsen


Data Science Ph.D. student at Worcester Polytechnic Institute.

View My GitHub Profile

I’m a Ph.D student in Data Science at Worcester Polytechnic Institute focusing on Machine Learning, advised by Prof. Elke Rundensteiner and collaborating with Prof. Xiangnan Kong and Cansu Sen in the Data Science Research Group.


I investigate sequential machine learning algorithms, particularly recurrent neural networks. My research revolves around three main topics related to recurrent neural networks: long-term dependencies in sequential data, asynchronous temporal data, and conditional computing. I also dabble in reinforcement learning, taking particular interest in gradient-based policy-search methods.

Please feel free to contact me with questions regarding my research or Data Science at WPI at


Current Research

I am currently interested in learning patterns in time series as early as possible. In many domains, the evaluation of a predictive model should not only be accuracy, but a trade-off between accuracy and earliness. In general I am interested in modeling very long sequences and sensor fusion through Recurrent Neural Networks. While so far my work has been evaluated in the clinical domain, I am far more interested in the theory behind these techniques, and these problems are present in many domains.