Structured learning modulo theories

DOI10.1016/J.ARTINT.2015.04.002zbMATH Open1404.68121arXiv1405.1675OpenAlexW2158797600MaRDI QIDQ511777FDOQ511777

Authors: Stefano Teso, Roberto Sebastiani, Andrea Passerini 

Publication date: 22 February 2017

Published in: Artificial Intelligence (Search for Journal in Brave)

Abstract: Modelling problems containing a mixture of Boolean and numerical variables is a long-standing interest of Artificial Intelligence. However, performing inference and learning in hybrid domains is a particularly daunting task. The ability to model this kind of domains is crucial in "learning to design" tasks, that is, learning applications where the goal is to learn from examples how to perform automatic {em de novo} design of novel objects. In this paper we present Structured Learning Modulo Theories, a max-margin approach for learning in hybrid domains based on Satisfiability Modulo Theories, which allows to combine Boolean reasoning and optimization over continuous linear arithmetical constraints. The main idea is to leverage a state-of-the-art generalized Satisfiability Modulo Theory solver for implementing the inference and separation oracles of Structured Output SVMs. We validate our method on artificial and real world scenarios.

Full work available at URL: https://arxiv.org/abs/1405.1675

Recommendations

• Learning Modulo Theories
• Model learning as a satisfiability modulo theories problem
• Learning relational dependency networks in hybrid domains
• Clustering-Guided SMT($$\mathcal {L\!R\!A}$$) Learning
• Learning to reason with a restricted view

zbMATH Keywords

satisfiability modulo theoryconstructive machine learninglearning with constraintsoptimization modulo theorystructured-output support vector machines

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Problem solving in the context of artificial intelligence (heuristics, search strategies, etc.) (68T20)

Cites Work

• Propagation of Probabilities, Means, and Variances in Mixed Graphical Association Models
• Title not available (Why is that?)
• Lazy satisfiability modulo theories
• Mixed integer programming computation
• Cutting-plane training of structural SVMs
• A Modular Approach to MaxSAT Modulo Theories
• On probabilistic inference by weighted model counting
• Title not available (Why is that?)
• Solving Non-linear Arithmetic
• Large margin methods for structured and interdependent output variables
• Bridging logic and kernel machines
• The magic of logical inference in probabilistic programming
• A hierarchy of relaxations for linear generalized disjunctive programming
• A Constraint Sequent Calculus for First-Order Logic with Linear Integer Arithmetic
• Title not available (Why is that?)
• Machine learning and knowledge discovery in databases. European conference, ECML PKDD 2011, Athens, Greece, September 5--9, 2011. Proceedings, Part II
• Satisfiability modulo the theory of costs: foundations and applications
• On SAT Modulo Theories and Optimization Problems
• Optimization in SMT with ${\mathcal LA}$ (ℚ) Cost Functions
• Optimization Modulo Theories with Linear Rational Costs
• Extending ProbLog with Continuous Distributions
• Compiling constraints in clp(FD)
• Stochastic local search for SMT: combining theory solvers with WalkSAT
• Minimal-Model-Guided Approaches to Solving Polynomial Constraints and Extensions
• Disjunctive Programming
• Computing layouts with deformable templates
• Inference in probabilistic logic programs with continuous random variables
• Pushing the Envelope of Optimization Modulo Theories with Linear-Arithmetic Cost Functions
• Model Evolution with Equality Modulo Built-in Theories
• Symbolic optimization with SMT solvers

Cited In (9)

• \textsc{OptiMathSAT}: a tool for optimization modulo theories
• Learning modulo theories for constructive preference elicitation
• Learning Modulo Theories
• Automatic synthesis of constraints from examples using mixed integer linear programming
• Delegatable Functional Signatures
• Ellipsoidal one-class constraint acquisition for quadratically constrained programming
• Learning relational dependency networks in hybrid domains
• Learning MAX-SAT from contextual examples for combinatorial optimisation
• Constraint learning: an appetizer

Uses Software

• ProbLog
• SYMBA
• Church
• MathSAT

This page was built for publication: Structured learning modulo theories

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q511777)