Marco Cusumano-Towner
Marco is a fourth-year Ph.D. student in electrical engineering and computer science at MIT, working with Vikash Mansinghka in the MIT Probabilistic Computing Project, and Josh Tenenbaum in the MIT Department of Brain and Cognitive Sciences.
Previously, Marco completed his Master's degree at Stanford University, where his research focused applied machine learning for computational biology. Marco has spent time in industry developing computational infrastructure and algorithms for genetic testing from high-throughput DNA sequencing data. During his undergraduate studies in at UC Berkeley, Marco worked with Professor Pieter Abbeel on probabilistic and optimization techniques for household robotics.
Marco is interested in developing programming languages, software systems, user interfaces, algorithms, and theory that make it easier to construct, reason about, and use probabilistic modeling and inference.
Session
This talk introduces a new flexible and extensible probabilistic programming system called Gen, that is built on top of Julia. Gen's extensible set of modeling DSLs can express probabilistic models that combine Bayesian networks, black box simulators, deep learning, structure learning, and Bayesian nonparametrics; and Gen's inference library supports custom algorithms that combine Markov chain Monte Carlo, particle filtering, variational inference, and numerical optimization.