AI Paper Atlas

Research map / Graph learning / GNN architectures

Beyond Simple Graphs: Neural Multi-Objective Routing on Multigraphs

Filip Rydin, Attila Lischka, Jiaming Wu, Morteza Haghir Chehreghani, Balazs Kulcsar

ICLR 2026 · Graph learning › GNN architectures · approach: Expressive equivariant message passing · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use GMS-EB instead of MBMMBM · MOCVRP XASY100normalized hypervolume: 0.60 -> 0.79Improved
use GMS-DH instead of MBMMBM · MGMOCVRP FLEX2-100normalized hypervolume: 0.81 -> 0.85Improved
use GMS-EB instead of MBMMBM · MGMOTSPTW FIX2-50normalized hypervolume: 0.64 -> 0.93Improved
add instance augmentation to GMS-EBGMS-EB · MOTSP TMAT50HV gap: 1.50% -> 1.14%Improved
add instance augmentation to GMS-EBGMS-EB · MOTSP TMAT100inference time: 3.6m -> 29mHurt
add pre-pruning before encoder to GMS-DH (ablation)GMS-DH · MGMOTSP FLEX2-50normalized hypervolume: 0.89 -> 0.77Hurt
add pre-pruning before encoder to GMS-DH (ablation)GMS-DH · MGMOCVRP FLEX2-50normalized hypervolume: 0.85 -> 0.75Hurt
add learned selection-decoder instead of simple linear pruning (ablation)GMS-DH Simple · MGMOTSPTW FLEX2-50normalized hypervolume: 0.83 -> 0.85Improved
add learned selection-decoder instead of simple linear pruning (ablation)GMS-DH Simple · MGMOTSPTW FIX2-50normalized hypervolume: 0.85 -> 0.91Improved
use GMS-DH instead of GMS-EBGMS-EB · MGMOTSP FLEX2-100inference time: 9.1m -> 49sImproved
use LKH instead of GMS-EBGMS-EB · MOTSP TMAT50HV gap: 1.50% -> 0.00%Mixed
use GMS-DH instead of GMS-EBGMS-EB · MOTSP TMAT200 (zero-shot)normalized hypervolume: 0.66 -> 0.65Mixed

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