AI Paper Atlas

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

Matrix encoding networks for neural combinatorial optimization

Yeong-Dae Kwon, Jinho Choo, Iljoo Yoon, Minah Park, Duwon Park, Youngjune Gwon

NeurIPS 2021 · Learning theory and algorithms › Combinatorial and search optimization · approach: Unsupervised neural combinatorial optimization · 16 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 instance augmentation (random one-hot assignments) instead of sampling at inference, x128 (ablation)MatNet · ATSP100gap to CPLEX: 1.89% -> 0.93%Improved
use instance augmentation x128 instead of sampling x128 (ablation)MatNet · ATSP20gap to CPLEX: 0.22% -> 0.01%Improved
increase sampling rollouts from 128 to 1280 (ablation)MatNet · ATSP100gap to CPLEX: 1.89% -> 1.58%Improved
use single shared update function for both groups (ablation)MatNet · ATSP100tour length: 1.621 -> 1.623No clear effect
use single shared update function for both groups (ablation)MatNet · FFSP100makespan: 91.529 -> 91.680Hurt
use x128 instance augmentation instead of single POMO rollout (ablation)MatNet · FFSP100makespan: 91.5 -> 89.7Improved
include graph embedding in QUERY token (ablation)MatNet · ATSPgapNo clear effect
replace one-hot initial embeddings with random vectors (ablation)MatNet · ATSP/FFSPsolution qualityHurt
MatNet vs best OR metaheuristic (PSO) on FFSP100MatNet · FFSP100makespan: 97.3 -> 91.5Improved

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