Machine learning (ML) has achieved many remarkable advances and successes in numerous areas. However, it is difficult for programmers to maintain, reuse, and extend ML programs. In particular, ML programs often come with a wide range of configurable options. As a result, programmers often need to define a family of programs. Currently they do this in an ad-hoc way, causing a significant amount of code duplication.
We study ML programming from a language design perspective. We propose a new choice-based learning paradigm which provides a separation of concerns that fosters modularity. The key insight underlying our design of choice-based learning is to combine two programming techniques: algebraic effects and handlers, and loss continuations. We establish the semantics of our design, implement our design as an effect handler library in Haskell, and provide various learning examples.
pdf (LAFI24.pdf) | 305KiB |
Sun 14 JanDisplayed time zone: London change
14:00 - 15:30 | Third SessionLAFI at Kelvin Lecture Chair(s): Steven Holtzen Northeastern University, Matthijs Vákár Utrecht University | ||
14:00 10mTalk | Effect Handlers for Choice-Based Learning LAFI File Attached | ||
14:10 10mTalk | Guaranteed Bounds for Discrete Probabilistic Programs with Loops via Generating Functions LAFI File Attached | ||
14:20 10mTalk | JuliaBUGS: A Graph-Based Probabilistic Programming Language using BUGS syntax LAFI Xianda Sun University of Cambridge, Philipp Gabler Independent researcher, Andrew Thomas University of Cambridge, Hong Ge University of Cambridge | ||
14:30 10mTalk | Mixture Languages LAFI File Attached | ||
14:40 10mTalk | Structured Tensor Algebra for Efficient Discrete Probabilistic Inference LAFI Amir Shaikhha University of Edinburgh | ||
14:50 10mTalk | Towards a Categorical Model of the Lilac Separation Logic LAFI John Li Northeastern University, Jon Aytac Sandia National Laboratories, Philip Johnson-Freyd Sandia National Laboratories, Amal Ahmed Northeastern University, USA, Steven Holtzen Northeastern University File Attached | ||
15:00 10mTalk | Toward Probabilistic Coarse-to-Fine Program Synthesis LAFI Maddy Bowers Massachusetts Institute of Technology, Alexander K. Lew Massachusetts Institute of Technology, Vikash K. Mansinghka Massachusetts Institute of Technology, Joshua B. Tenenbaum Massachusetts Institute of Technology, Armando Solar-Lezama Massachusetts Institute of Technology | ||
15:10 10mTalk | Static Posterior Inference of Bayesian Probabilistic Programming via Polynomial SolvingOnline LAFI Peixin Wang University of Oxford, Hongfei Fu Shanghai Jiao Tong University, Tengshun Yang SKLCS, Institute of Software, Chinese Academy of Sciences & University of Chinese Academy of Sciences, Guanyan Li University of Oxford, C.-H. Luke Ong NTU | ||
15:20 10mTalk | Abstract Interpretation for Automatic DifferentiationOnline LAFI Jacob Laurel University of Illinois at Urbana-Champaign, Siyuan Brant Qian University of Illinois at Urbana-Champaign; Zhejiang University, Gagandeep Singh University of Illinois at Urbana-Champaign; VMware Research, Sasa Misailovic University of Illinois at Urbana-Champaign |