Convex Compositional Reasoning Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Roketlishvili, Meir, Semenov, Semyon, Bobrin, Maksim, Kovalchuk, Viktor, Baichorov, Albert, Shtanchaev, Abduragim, Karray, Fakhri, Dylov, Dmitry V., Takáč, Martin, Asadulaev, Arip
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913161094889472
author Roketlishvili, Meir
Semenov, Semyon
Bobrin, Maksim
Kovalchuk, Viktor
Baichorov, Albert
Shtanchaev, Abduragim
Karray, Fakhri
Dylov, Dmitry V.
Takáč, Martin
Asadulaev, Arip
author_facet Roketlishvili, Meir
Semenov, Semyon
Bobrin, Maksim
Kovalchuk, Viktor
Baichorov, Albert
Shtanchaev, Abduragim
Karray, Fakhri
Dylov, Dmitry V.
Takáč, Martin
Asadulaev, Arip
contents Compositional energy-based models can generalize to larger combinatorial reasoning problems by reusing a learned factor energy across many local constraints. In our paper, we show that a key bottleneck in compositional reasoning is not composition itself, but the non-convex geometry of the learned energy landscape. To solve this problem, we introduce Convex Compositional Energy Minimization (CCEM), a framework that parameterizes each factor with an input-convex neural network and optimizes the composed energy over a tight convex relaxation of the feasible set. Because convexity is preserved under summation, the global relaxed objective remains convex, enabling deterministic projected first-order optimization. CCEM is trained in two stages: factor-level contrastive learning to shape local energy basins, followed by end-to-end refinement through an unrolled projected solver. Our experiments show that our models trained on small subproblems or a single problem size transfer to larger instances without retraining.
format Preprint
id arxiv_https___arxiv_org_abs_2605_23395
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Convex Compositional Reasoning Models
Roketlishvili, Meir
Semenov, Semyon
Bobrin, Maksim
Kovalchuk, Viktor
Baichorov, Albert
Shtanchaev, Abduragim
Karray, Fakhri
Dylov, Dmitry V.
Takáč, Martin
Asadulaev, Arip
Machine Learning
Compositional energy-based models can generalize to larger combinatorial reasoning problems by reusing a learned factor energy across many local constraints. In our paper, we show that a key bottleneck in compositional reasoning is not composition itself, but the non-convex geometry of the learned energy landscape. To solve this problem, we introduce Convex Compositional Energy Minimization (CCEM), a framework that parameterizes each factor with an input-convex neural network and optimizes the composed energy over a tight convex relaxation of the feasible set. Because convexity is preserved under summation, the global relaxed objective remains convex, enabling deterministic projected first-order optimization. CCEM is trained in two stages: factor-level contrastive learning to shape local energy basins, followed by end-to-end refinement through an unrolled projected solver. Our experiments show that our models trained on small subproblems or a single problem size transfer to larger instances without retraining.
title Convex Compositional Reasoning Models
topic Machine Learning
url https://arxiv.org/abs/2605.23395