Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Defresne, Marianne, Gambardella, Romain, Barbe, Sophie, Schiex, Thomas
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908719425519616
author Defresne, Marianne
Gambardella, Romain
Barbe, Sophie
Schiex, Thomas
author_facet Defresne, Marianne
Gambardella, Romain
Barbe, Sophie
Schiex, Thomas
contents In the ongoing quest for hybridizing discrete reasoning with neural nets, there is an increasing interest in neural architectures that can learn how to solve discrete reasoning or optimization problems from natural inputs, a task that Large Language Models seem to struggle with. Objectives: We introduce a differentiable neuro-symbolic architecture and a loss function dedicated to learning how to solve NP-hard reasoning problems. Methods: Our new probabilistic loss allows for learning both the constraints and the objective, thus delivering a complete model that can be scrutinized and completed with side constraints. By pushing the combinatorial solver out of the training loop, our architecture also offers scalable training while exact inference gives access to maximum accuracy. Results: We empirically show that it can efficiently learn how to solve NP-hard reasoning problems from natural inputs. On three variants of the Sudoku benchmark -- symbolic, visual, and many-solution --, our approach requires a fraction of training time of other hybrid methods. On a visual Min-Cut/Max-cut task, it optimizes the regret better than a Decision-Focused-Learning regret-dedicated loss. Finally, it efficiently learns the energy optimization formulation of the large real-world problem of designing proteins.
format Preprint
id arxiv_https___arxiv_org_abs_2508_20978
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives
Defresne, Marianne
Gambardella, Romain
Barbe, Sophie
Schiex, Thomas
Artificial Intelligence
Logic in Computer Science
Symbolic Computation
In the ongoing quest for hybridizing discrete reasoning with neural nets, there is an increasing interest in neural architectures that can learn how to solve discrete reasoning or optimization problems from natural inputs, a task that Large Language Models seem to struggle with. Objectives: We introduce a differentiable neuro-symbolic architecture and a loss function dedicated to learning how to solve NP-hard reasoning problems. Methods: Our new probabilistic loss allows for learning both the constraints and the objective, thus delivering a complete model that can be scrutinized and completed with side constraints. By pushing the combinatorial solver out of the training loop, our architecture also offers scalable training while exact inference gives access to maximum accuracy. Results: We empirically show that it can efficiently learn how to solve NP-hard reasoning problems from natural inputs. On three variants of the Sudoku benchmark -- symbolic, visual, and many-solution --, our approach requires a fraction of training time of other hybrid methods. On a visual Min-Cut/Max-cut task, it optimizes the regret better than a Decision-Focused-Learning regret-dedicated loss. Finally, it efficiently learns the energy optimization formulation of the large real-world problem of designing proteins.
title Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives
topic Artificial Intelligence
Logic in Computer Science
Symbolic Computation
url https://arxiv.org/abs/2508.20978