Massachusetts Institute of Technology MIT CSAIL MIT BCS


We build probabilistic computing systems that exploit uncertain knowledge to learn from data, infer its probable causes, make calibrated predictions and choose effective actions. We also study the computational principles and building blocks needed to design, implement and analyze these systems, drawing on and contributing to an emerging integration of key ideas from probability theory and computer science. Our research includes work on machine learning and artificial intelligence fundamentals, as well as applications to modeling human cognition and to intelligent data analysis.

So far, this work has yielded new general-purpose probabilistic programming technology and intentionally stochastic (but still digital) hardware for real-time Bayesian inference. It has also yielded academic and commercial Bayesian database systems that automate the analysis of high-dimensional data tables.


  • Venture --- an interactive, Turing-complete probabilistic programming platform descended from Church
  • BayesDB --- a Bayesian database that lets users query the probable implications of their data, and solve basic data science problems without training in statistics
  • Picture --- an imperative probabilistic programming language for scene perception, building on the GPGP framework
  • CrossCat --- a nonparametric Bayesian machine learning method for analyzing high-dimensional data tables

Current Members

  • Taylor Campbell (Research Engineer)
  • Baxter Eaves (Research Engineer)
  • Vlad Firoiu (PhD Student, EECS; formerly undergraduate)
  • Tejas Kulkarni (PhD Student, BCS)
  • Vikash Mansinghka (Research Scientist, PI)
  • Alexey Radul (Research Engineer)
  • David Wadden (Research Assistant)
  • Benjamin Zinberg (MEng Student, EECS)

Former Members

  • Dan Lovell (Research Engineer) -> software engineer at Continuum Analytics (2013)
  • Daniel Selsam (Research Assistant) -> graduate student at Stanford (2014)
  • Jay Baxter (MEng Student, EECS) - data scientist at Twitter
  • Yura Perov (Visiting Undergraduate) -> visiting undergraduate at Oxford (2013)
  • Jeff Wu (MEng Student, EECS) -> software engineer #1 at (2012)
  • Owain Evans (PhD Student, Philosophy) -> postdoctoral fellow at Oxford (2015)


Most recent

  • Querying the probable implications of tabular data with BQL and BayesDB. Mansinghka, V., Baxter, J., Eaves, B. [extended abstract presented at 3rd NIPS Workshop on Probabilistic Programming], 2014.
  • Venture: a general purpose probabilistic programming platform with efficient stochastic inference. Mansinghka, V., Radul, A. [pre­print available as arXiv, number 1404:0099] [pdf]
  • Discrete Particle Variational Inference. Saeedi, A., Kulkarni, T., Mansinghka, V., Gershman, [extended abstract delivered as oral presentation at 4th NIPS Workshop on Advances in Variational Inference], 2014.
  • CoreVenture: a high­level, reflective machine language for probabilistic programming. Mansinghka, V., Radul, A. [extended abstract presented at 3rd NIPS Workshop on Probabilistic Programming], 2014.
  • VentureScript: a language for probabilistic modeling and interactive inference. Mansinghka, V., Radul, A. [extended abstract presented at 3rd NIPS Workshop on Probabilistic Programming], 2014.
  • Picture: an imperative probabilistic programming language for scene perception. Kulkarni, T., Kohli, P., Tenenbaum, J., Mansinghka, V. [Under review at CVPR.] [pdf] [supplemental]
  • Sublinear approximate inference for probabilistic programs, Chen, Y., Mansinghka, V., Ghahramani, Z. [Under review at AISTATS; pre­print available as arXiv, number 1411.1690, 2014.] [pdf]
  • Particle Gibbs with Ancestor Resampling for Probabilistic Programs. Willem, J., Yeung, H., Wood, F., and Mansinghka, V. [Accepted to AISTATS 2015; pre­print available as arXiv, number 1501.06769] [pdf]
  • JUMP Means: Small Variance Asymptotics for Markov Jump Processes. Huggins, J., Narasimhan, K., Saeedi, A., and Mansinghka, V. [Under review at ICML.]
  • Building fast Bayesian computing machines out of intentionally stochastic parts. Mansinghka, V., Jonas, E. [pre­print available as arXiv, number 1402:4914], 2014. [pdf]

  • Composable gradient­based inference for higher­order probabilistic programming languages. Mansinghka, V., Radul, A. [Presented at 3rd NIPS Workshop on Probabilistic Programming.], 2014.
  • GPGP3D: Generative probabilistic graphics programming for 3D shape perception. Kulkarni, T., Kohl, P., Tenenbaum, J., Mansinghka, V. [Presented at 3rd NIPS Workshop on Probabilistic Programming.], 2014.
  • plotf: a candidate printf for probabilistic programs. Mansinghka, V., Radul, A. [Presented at 3rd NIPS Workshop on Probabilistic Programming.], 2014.
  • Identifying speed and convergence bottlenecks in probabilistic programs. Firoiu, V., Zinberg, B., Mansinghka, V. [3rd NIPS Workshop on Probabilistic Programming.], 2014.
  • Writing custom proposals for probabilistic programs as probabilistic programs. Zhang, H., Mansinghka, V. [Presented at 3rd NIPS Workshop on Probabilistic Programming.], 2014.
  • A new approach to probabilistic programming inference. Wood, F., van den Meent, J., Mansinghka, V., AISTATS, 2014. [Oral]. [pdf]
  • Automatic inference for inverting software simulators via probabilistic programming. Saeedi, A., Firoiu, V., Mansinghka, V. [ICML Workshop on Automatic Machine Learning.], 2014. [pdf]

  • Markov chain algorithms: A template for building future robust low power systems. Deka, B., Birklykke, A., Duwe, H., Mansinghka, V., Kumar, R. Asilomar Conference on Signals, Systems, and Computers, 2013. [pdf]
  • Approximate Bayesian image interpretation using generative probabilistic graphics programs. Mansinghka, V., Kulkarni, T., Perov, Y., Tenenbaum, J. NIPS, 2013. [Oral] [pdf]
  • Parallel Markov chain Monte Carlo for Dirichlet process mixtures. Lovell, D., Malmaud, J., Adams, R. P., Mansinghka, V. [Oral presentation at NIPS Workshop on Big Learning.], 2012. Extended version available on arXiv, number 1304.2302, 2013. [pdf]
  • Reconciling intuitive physics and Newtonian mechanics for colliding objects. Sanborn, A., Mansinghka, V., and Griffiths, T.L. [Psychological Review], 2013. [pdf]
2011 and earlier

  • A probabilistic model of cross­categorization. Shafto, P., Kemp, C., Mansinghka, V., Tenenbaum, J. B. [Cognition], 2011. [pdf]
  • Beyond calculation: probabilistic computing machines and universal stochastic inference. Mansinghka, V. [NIPS Workshop on Philosophy and Machine Learning.], 2011. [Oral]
  • When are probabilistic programs probably computationally tractable? Freer, C., Mansinghka, V., Roy, D. [NIPS Workshop on Monte Carlo Methods for Modern Applications.], 2010. [pdf]
  • Cross categorization: a method for discovering multiple overlapping clusterings. Mansinghka, V., Jonas, E., Petschulat, C., Cronin, B., Shafto, P. and Tenenbaum, J. [NIPS Workshop on Nonparametric Bayes.], 2009. [pdf]
  • A stochastic programming perspective on nonparametric Bayes. Roy, Mansinghka, Goodman and Tenenbaum. [ICML Nonparametric Bayes Workshop.], 2008. [Oral] [pdf]
  • CrossCat: a fully Bayesian, nonparametric method for analyzing heterogeneous, high- dimensional data. Mansinghka, V., Shafto, P. Jonas, E., Petschulat, C., Gasner, M. and Tenenbaum, J. [Journal of Machine Learning Research, accepted pending revision.]
  • A Bayesian framework for modeling intuitive dynamics. Sanborn, A., Mansinghka, V., Griffiths, T. L. COGSCI, 2009. [pdf]
  • Exact and approximate sampling by systematic stochastic search. Mansinghka, V., Roy, D., Jonas, E., Tenenbaum, J. B. AISTATS, 2009. [Oral]. [pdf]
  • Natively Probabilistic Computation. Mansinghka, V. MIT Doctoral dissertation. 2009. Winner, George M. Sprowls dissertation award. [pdf]
  • Nonparametric Bayesian Models for Supervised and Unsupervised Learning. Mansinghka, V. MIT Masters thesis. 2009. [link]
  • Stochastic Digital Circuits for Probabilistic Inference. Mansinghka, V., Jonas, E., Tenenbaum, J. B. MIT CSAIL Technical Report, 2008. [pdf]
  • Church: a universal language for generative models. Goodman, N. and Mansinghka, V., Roy, D., Bonawitz, K., Tenenbaum, J. B. UAI, 2008. [Oral, available as arXiv, number 1206.3255]. [pdf]
  • Learning grounded causal models. Goodman, N., Mansinghka, V., Tenenbaum, J. B. COGSCI, 2007. [Oral; winner of Cognitive Science Society Computational Modeling Prize for Perception and Action] [pdf]
  • Modeling human performance in statistical word segmentation. Frank, M. C., Goldwater, S., Mansinghka, V., Griffiths, T., Tenenbaum, J. B. COGSCI, 2007. [pdf]
  • AClass: an online algorithm for generative classification. Mansinghka, V. Roy, D. Rifkin, R. Tenenbaum, J. B. AISTATS, 2007. [pdf]
  • Learning annotated hierarchies from relational data. Roy, D., Kemp, C., Mansinghka, V. K., and Tenenbaum, J. B. NIPS, 2006. [Oral] [pdf]
  • Structured priors for structure learning. Mansinghka, V. K., Kemp, C., Tenenbaum, J. B., and Griffiths, T. L. UAI, 2006. [Oral] [pdf]
  • Learning cross­cutting systems of categories. Shafto, P. Kemp, C., Mansinghka, V., Gordon, M., and Tenenbaum, J. B. COGSCI, 2006. [Poster].
  • Intuitive theories of mind: A rational approach to false belief. Goodman, N., Baker, C., Bonawitz­Baraff, E., Mansinghka, V., Gopnik, A., Wellman, H., Shulz, L., Tenenbaum, J. B. COGSCI, 2006. [pdf]


The MIT Probabilistic Computing Project is hosted by MIT's Computer Science and Artificial Intelligence Laboratory and Department of Brain and Cognitive Sciences. Our work is generously supported by research contracts with DARPA (under the XDATA and PPAML programs), the Office of Naval Research and the Army Research Laboratory, and Shell Oil, as well as gifts from Analog Devices and Google. The views expressed on this website and in our research are our own, and do not necessarily reflect the views of our government or corporate sponsors.