AI Paper Atlas

Research map / Learning theory and algorithms / Combinatorial and search optimization

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

Niki Triantafyllou, Maria Papathanasiou

NeurIPS 2026 · Learning theory and algorithms › Combinatorial and search optimization · approach: Unsupervised neural combinatorial optimization · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes EnerGNN, a learned energy over binary configurations trained with optimality-aware contrastive learning so that a MIP layer can enforce hard constraints exactly at inference.

Problem

Constrained combinatorial inference where hard constraints and deployment-time restrictions must be satisfied exactly, which learned models typically fail to guarantee.

Approach

Learn an energy function over binary configurations that is compatible with a mixed-integer programming layer, shaping the energy landscape with optimality-aware contrastive learning so that constraints are enforced exactly at inference time.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace linear energy head with ReLU MLP head (ablation)EnerGNN · QKPmean primal gap (%): 26.87 -> 0.31Improved
replace linear energy head with ReLU MLP head (ablation)EnerGNN · CARTmean primal gap (%): 21.80 -> 0.05Improved
replace linear energy head with ReLU MLP head (ablation)EnerGNN · CAmean primal gap (%): 0.20 -> 0.53Hurt
add calibration loss (ablation)EnerGNN · CARTmean primal gap (%): 0.78 -> 0.05Improved
add calibration loss (ablation)EnerGNN · CAmean primal gap (%): 7.69 -> 0.53Improved
add calibration loss (ablation)EnerGNN · QKPinstances within 1% (%): 68 -> 99Improved
add infeasibility loss (ablation)EnerGNN · CART, QKP, CAmean primal gap (%): 0.06 -> 0.05 (CART)No clear effect
optimize learned energy under MIP instead of restricting search (vs ConPaS)EnerGNN · CARTtime (s): 3404.19 -> 214.03Improved
transfer trained model to different size without retrainingEnerGNN · CART m=16 to 12 and 20mean primal gap (%): 0.05 -> 0.98 (m=12), 1.23 (m=20)Hurt
increase CA bids from 175 to 250 at test time without retrainingEnerGNN (Linear) · CAmean primal gap (%): 0.20 -> 0.98Hurt
zero-shot constraint injection (deactivate 3 facilities)EnerGNN · CARTmean primal gap (%): 0.21 -> 0.45 (remove 1 to 3)Hurt
zero-shot constraint injection vs Neural DivingNeural Diving · CART remove 3 facilitiesfeasibility (%): 49.12 feasibilityHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy