Logic meets Learning
Tutorial to be held at ESSAI 2026, Vienna
Abstract
The tension between deduction and induction remains fundamental to artificial intelligence. This tutorial surveys the intersection of logic and learning, exploring how these historically distinct paradigms can be unified. We examine three strands: logic versus learning (including weighted model counting and knowledge compilation), machine learning for logic (inductive logic programming, Bayesian scoring, PAC-semantics), and logic for machine learning (probabilistic programming, algebraic model counting, abstraction). The tutorial emphasizes both theoretical foundations and practical algorithmic techniques, demonstrating how formal languages can capture knowledge while learning generalizes from examples. Attendees will gain understanding of statistical relational learning, neuro-symbolic systems, and the mathematical frameworks connecting symbolic reasoning with data-driven approaches. The material bridges classical AI and modern machine learning, preparing researchers for cross-over applications in high-level control, knowledge representation, and automated reasoning.
Course Contents
The course consists of 4 lectures (4 x 90 min):
Lecture 1: Foundations and Logic vs Learning
Motivation: the deduction-induction divide in AI. Background: propositional and first-order logic, probabilistic models. Logic vs learning: weighted model counting.
Lecture 2: Machine Learning for Logic
Inductive logic programming: learning rules from examples. Case study: neuro-symbolic rule learning via differential inductive logic programming.
Lecture 3: Logic for Machine Learning
Probabilistic programming languages. Algebraic model counting. Abstraction. Case study: multilinear models for diverse counterfactuals.
Lecture 4: Integration and Advanced Topics
Neuro-symbolic systems: combining neural networks with symbolic reasoning. LLMs as symbolic executors. Open challenges and future directions.
Who is this for
The course is pitched at an introductory level suitable for graduate students, researchers, and practitioners. It assumes basic knowledge of propositional and first-order logic and familiarity with probability theory. No prior knowledge of statistical relational learning or neuro-symbolic AI is assumed. Cf chapter to see contents covered.
Previous Venues
This tutorial is a significantly expanded and revised version of ones previously delivered at IJCAI (2017), KR (2022), and the Logic in AI Summer School, Como (2024).