Dyna – Logic Programming for Machine Learning

Dyna is a programming language for machine learning research, building on logic programming but adding flexible execution and weighted rules. It allows complex algorithms to be expressed concisely, bridging mathematical concepts and executable programs. Dyna evolved from version 1.0, which extended Datalog, to 2.0, which added functions, lazy expressions, and prototype-based inheritance. Current research aims to improve implementation and optimize execution through reinforcement learning.