← Back to all news
arXiv Machine Learning By Niki Triantafyllou, Maria M. Papathanasiou

EnerGNN: Learning Optimization-Compatible Energy Functions for Exact Constrained Combinatorial Inference

Read the original on arXiv Machine Learning →

EnerGNN is a framework that learns energy functions over complete binary configurations and performs exact constrained inference via mixed integer programming. It combines a graph neural network encoder, an optimization‑compatible readout, and a nonlinear energy head, and is trained with an optimality‑aware contrastive loss to reflect solution quality and constraint violations. In experiments on Quadratic Knapsack, Combinatorial Auctions, and a personalized cell therapy supply chain problem, EnerGNN achieves near‑optimal solutions with 100% feasibility, a 0.05% mean primal gap, and a 17.4× speedup over a monolithic MIP, while also generalizing to larger instances and supporting zero‑shot constraint injection.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Sep 23

Deep Reinforcement Learning on Item-Compatibility Graphs for One-Dimensional Bin Packing

The paper introduces a novel end‑to‑end, size‑agnostic graph reinforcement learning framework for the one‑dimensional bin packing problem (1D‑BPP). It models packing as a Markov decision process on an item‑compatibility graph, where a graph neural network actor‑critic policy learns to merge compatible partial bins. Empirical results on the BPPLIB benchmark show that the learned policy reduces the mean optimality gap of a constructive heuristic from 2.66 % to 2.31 %, performs competitively against other learned methods, and outperforms a state‑of‑the‑art learned solver on the hardest benchmark family.

By M. Asl{\i} Ayd{\i}n
reinforcement-learningbenchmarks
More like this →
arXiv Machine Learning
Oct 3

Reformulation-Contrastive Learning for Mixed Integer Programs

The paper introduces ReMILP, a reformulation‑contrastive learning framework that uses self‑supervision from equivalent formulations of mixed‑integer linear programs (MILPs). By distinguishing re‑descriptions and substitutions, the method trains a graph neural network and a hypernetwork to predict how variable embeddings transform under changes of variables, achieving invariance and equivariance without solver‑derived labels. The learned representations prove useful for tasks such as binary solution, constraint activity, and integrality gap prediction, and serve as a strong initialization for fine‑tuning.

By Ousema Bouaneni, Mathis Le Bail, Cl\'ement Elliker, Ma\"el Jenny, Sonia Vanier
ragfine-tuning
More like this →
arXiv AI
Sep 17

Variational Approach for Job Shop Scheduling

The paper introduces Variational Graph-to-Scheduler (VG2S), a framework that applies variational inference to the Job Shop Scheduling Problem (JSSP). By decoupling representation learning from policy optimization using a variational graph encoder and an ELBO-based objective, VG2S improves training stability and robustness to hyperparameter changes. Experiments show that VG2S outperforms state‑of‑the‑art deep reinforcement learning baselines and traditional dispatching rules, especially on large‑scale benchmark instances such as DMU and SWV.

By Seung Heon Oh, Jiwon Baek, Hyunjin Oh, Kiyoung Cho, Heechang Yoon, Jong Hun Woo
agentsreinforcement-learningbenchmarks
More like this →