Marco Cusumano-Towner

Email / GitHub / Google Scholar

Computer scientist specializing in probabilistic AI.

I am a research scientist at Apple working with Vladlen Koltun. I completed my PhD in EECS at MIT, where I was advised by Vikash Mansinghka and Josh Tenenbaum. During my PhD I created the Gen probabilistic programming system. Before MIT, I was a technical lead at an early stage molecular diagnostics startup backed by Sequoia Capital, I completed my MS in computer science at Stanford University and my BS in electrical engineeering and computer science at UC Berkeley, where I worked with Pieter Abbeel on probabilistic robotics. My graduate school research was funded by the NSF graduate research fellowship and the NDSEG graduate fellowship program, among other sources.

PhD Thesis

Gen: A High-Level Programming Platform for Probabilistic Inference (PDF), PhD thesis, Massachusetts Institute of Technology, 2020.

Publications

Lew, Cusumano-Towner, Mansinghka Recursive Monte Carlo and Variational Inference with Auxiliary Variables The 38th Conference on Uncertainty in Artificial Intelligence (UAI 2022)

Saad, Cusumano-Towner, Mansinghka Interval Estimators of Entropy and Information Measures via Inference in Probabilistic Models The 25th International Conference on Artificial Intelligence and Statistics (AISTATS 2022)

Gothoskar, Cusumano-Towner, Zinberg, Ghavamizadeh, Pollok, Garrett, Gutfreund, Tenenbaum, Mansinghka 3DP3: 3D Scene Perception via Probabilistic Programming Thirty-first Conference on Neural Information Processing Systems (NeurIPS 2021)

Cusumano-Towner, Lew, Mansinghka Automating Involutive MCMC using Probabilistic and Differentiable Programming. arXiv 2020

Lew, Cusumano-Towner, Sherman, Carbin, Mansinghka. Trace types and denotational semantics for sound programmable inference in probabilistic languages. 47th ACM SIGPLAN Symposium on Principles of Programming Languages (POPL 2020)

Cusumano-Towner, Saad, Lew, Mansinghka. Gen: a general-purpose probabilistic programming system with programmable inference. 40th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2019)

Saad, Cusumano-Towner, Schaechtle, Rinard, Mansinghka Bayesian synthesis of probabilistic programs for automatic data modeling. 46th ACM SIGPLAN Symposium on Principles of Programming Languages (POPL 2019)

Zinberg, Cusumano-Towner, Mansinghka Structured differentiable models of 3D scenes via generative scene graphs. NeurIPS 2019 Workshop on Perception as Generative Reasoning.

Cusumano-Towner, Bichsel, Gehr, Vechev, Mansinghka Incremental inference for probabilistic programs. 39th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI 2018)

Cusumano-Towner, Mansinghka Using probabilistic programs as proposals arXiv 2018

Cusumano-Towner, Mansinghka AIDE: An algorithm for measuring the accuracy of probabilistic inference algorithms Thirty-first Conference on Neural Information Processing Systems (NeurIPS 2017)

Cusumano-Towner, Radul, Wingate, Mansinghka Probabilistic programs for inferring the goals of autonomous agents arXiv 2017

Cusumano-Towner, Mansinghka Encapsulating models and approximate inference programs in probabilistic modules arXiv 2017

Cusumano-Towner, Li, Tuo, Krishnan, Maslove A social network of hospital acquired infection built from electronic medical record data Journal of the American Medical Informatics Association. 2013.

Cusumano-Towner, Singh, Miller, O'Brien, Abbeel Bringing clothing into desired configurations with limited perception. 2011 IEEE International Conference on Robotics and Automation (ICRA 2011)

Maitin-Shepard, Cusumano-Towner, Lei, Abbeel Cloth grasp point detection based on multiple-view geometric cues with application to robotic towel folding. 2010 IEEE International Conference on Robotics and Automation (ICRA 2010)

News