Effect Handlers for Choice-Based Learning in Machine Learning Programming
Explore a groundbreaking approach to machine learning programming in this 14-minute conference talk from ACM SIGPLAN's LAFI'24. Delve into the concept of choice-based learning, a paradigm designed to enhance modularity and reduce code duplication in ML programs. Discover how the combination of algebraic effects, handlers, and loss continuations can revolutionize the way programmers maintain, reuse, and extend machine learning code. Learn about the semantics behind this innovative design and its implementation as an effect handler library in Haskell. Gain insights into various learning examples that demonstrate the practical applications of this new paradigm, potentially transforming the landscape of ML programming for improved efficiency and flexibility.