<div dir="ltr"><div><span style="font-family:monospace">tl;dr:<br>* List of talks and abstracts is out.<br>* Early registration deadline is Wednesday 18 Dec (AoE, TODAY).</span></div><div><span style="font-family:monospace"><br></span></div><div><span style="font-family:monospace">LAFI 2020: Languages for Inference (formerly PPS)</span></div><div><span style="font-family:monospace"></span></div><span style="font-family:monospace">================================================<br>Tuesday, 21 January 2020, New Orleans, Louisiana, US<br>A workshop affiliated with POPL 2020<br><a href="https://popl20.sigplan.org/home/lafi-2020">https://popl20.sigplan.org/home/lafi-2020</a><br><br>Important dates (anywhere on earth)<br>-------------------------------------------------<br> Early registration deadline    Wed 18 Dec 2019 (TODAY)<br> Workshop                       Tue 21 Jan 2020<br>-------------------------------------------------<br><br>Registration: <a href="https://popl20.sigplan.org/attending/Registration">https://popl20.sigplan.org/attending/Registration</a><br><br>Invited speaker: Fritz Obermeyer, Uber AI Labs<br><br>    Nonstandard Interpretation in Pyro<br><br><a href="https://popl20.sigplan.org/details/lafi-2020/1/Invited-talk-Nonstandard-Interpretation-in-Pyro">https://popl20.sigplan.org/details/lafi-2020/1/Invited-talk-Nonstandard-Interpretation-in-Pyro</a><br><br>Accepted talks: <a href="https://popl20.sigplan.org/home/lafi-2020#event-overview">https://popl20.sigplan.org/home/lafi-2020#event-overview</a><br><br>Context<br>=======<br><br>Inference concerns re-calibrating program parameters based on<br>observed data, and has gained wide traction in machine learning and<br>data science. Inference can be driven by probabilistic analysis and<br>simulation, and through back-propagation and<br>differentiation. Languages for inference offer built-in support for<br>expressing probabilistic models and inference methods as programs, to<br>ease reasoning, use, and reuse. The recent rise of practical<br>implementations as well as research activity in inference-based<br>programming has renewed the need for semantics to help us share<br>insights and innovations.<br><br>This workshop aims to bring programming-language and machine-learning<br>researchers together to advance all aspects of languages for<br>inference. Topics include but are not limited to:<br><br>+ design of programming languages for inference and/or differentiable<br>  programming;<br>+ inference algorithms for probabilistic programming languages,<br>  including ones that incorporate automatic differentiation;<br>+ automatic differentiation algorithms for differentiable programming<br>  languages;<br>+ probabilistic generative modelling and inference;<br>+ variational and differential modelling and inference;<br>+ semantics (axiomatic, operational, denotational, games, etc) and<br>  types for inference and/or differentiable programming;<br>+ efficient and correct implementation;<br>+ and last but not least, applications of inference and/or<br>  differentiable programming.<br><br>For a sense of the talks, posters, and blogs in past years, see<br><br>+ LAFI-2019: <a href="https://popl20.sigplan.org/track/lafi-2019">https://popl20.sigplan.org/track/lafi-2019</a><br><br>+ PPS-2018: <a href="http://conf.researchr.org/track/POPL-2018/pps-2018">http://conf.researchr.org/track/POPL-2018/pps-2018</a><br>  blog:     <a href="http://pps2018.soic.indiana.edu/">http://pps2018.soic.indiana.edu/</a><br><br>+ PPS-2017: <a href="http://conf.researchr.org/track/POPL-2017/pps-2017">http://conf.researchr.org/track/POPL-2017/pps-2017</a><br>  blog:     <a href="http://pps2017.soic.indiana.edu/">http://pps2017.soic.indiana.edu/</a>)<br><br>+ PPS-2016: <a href="http://conf.researchr.org/track/POPL-2016/pps-2016">http://conf.researchr.org/track/POPL-2016/pps-2016</a><br>  blog:     <a href="http://pps2016.soic.indiana.edu/">http://pps2016.soic.indiana.edu/</a>)<br><br>Last year we explicitly expanded the focus of the workshop from<br>statistical probabilistic programming to encompass differentiable<br>programming for statistical machine learning. This change seemed<br>well-received by the community, and we continue it this year<br>in an effort to extend the strong ties between programming<br>language-based machine learning and the POPL community.<br><br>We expect this workshop to be informal, and our goal is to foster<br>collaboration and establish common ground involving ongoing work on<br>probabilistic and differentiable programming languages, semantics, and<br>systems.<br><br><br>Programme committee:<br>Justin Hsu,                   University of Wisconsin-Madison, USA<br>Ohad Kammar (co-chair)        University of Edinburgh, UK (co-chair)<br>Jerzy Karczmarczuk            France<br>Marie Kerjean                 Inria Nantes, France<br>Dougal Maclaurin (co-chair)   Google Brain, USA (co-chair)<br>Barak A. Pearlmutter          Maynooth University, Ireland<br>David Tolpin                  PUB+, Israel<br>Andrea Walther                Humboldt-Universität zu Berlin, Germany<br>Richard Wei                   Apple Inc., USA<br></span></div>