AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Graph distance and retrieval

Learn graph edit distance and similarity for graph retrieval and coarsening.

79 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Graph algorithms and clustering, keyword cluster “search · directed · causal”

State History Tried Baselines Checklist Preprints Who Export

Where it stands

Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.

25 papers in 2023–24 and 17 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 12.8% to 8.5%.

Open questions the data points to

How this is decided

A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.

How it got here

2016: 3162017: 2172018: 4182019: 7192020: 6202021: 7212022: 8222023: 11232024: 14242025: 9252026: 826

Most cited paper of each year

Earliest papers here

What has been tried

64 papers here report 606 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 8%, hurt 17%, mixed 6%. Papers mostly publish changes that helped, so "helped" is over-represented.

Ablations in recent papers

What was changedMeasured onResultEffectPaper
Use BFS serialization instead of DFSnuScenes · M-F73.3 -> 72.1HurtLaneGraph2Seq: Lane Topology Extraction with Language Model via Vertex-Edge Encoding and Connectivity Enhancement (AAAI 2024 · 10 citations)
Add local FP resamplingMNIST · cluster separationHelpedDimension Reduction with Locally Adjusted Graphs (AAAI 2025 · 6 citations)
Add the subset mechanism to TREE-GAll 17 graph and node prediction tasks · accuracy or AUC (Country: RMSE)HelpedTREE-G: Decision Trees Contesting Graph Neural Networks (AAAI 2024 · 4 citations)
Use RGGC instead of GIN or GCN as the backboneAIDS700nef GED · MAE0.677 ±0.013 -> 0.640 ±0.013; 0.654 ±0.011 -> 0.640 ±0.013HelpedGraSP: Simple Yet Effective Graph Similarity Predictions (AAAI 2025 · 3 citations)
add the GRU convolution instead of using GIN convolutionRome · stress238.53 ± 1.43 -> 233.17 ± 0.13HelpedCoRe-GD: A Hierarchical Framework for Scalable Graph Visualization with GNNs (ICLR 2024 · 2 citations)
Add trainable representations for non-edges and use all node-pairsMutag, Code2, and Molhiv, uniform cost · MSE0.566 -> 0.492; 0.683 -> 0.429; 0.858 -> 0.781HelpedGraph Edit Distance with General Costs Using Neural Set Divergence (NeurIPS 2024 · 2 citations)
Set Δ0=Δ2=10 rather than Δ0=Δ2=0 (Δ1=0)baboons · closure number, weak closure number8, 5 -> 5, 3HelpedTemporal Triadic Closure: Finding Dense Substructures in Social Networks That Evolve over Time (AAAI 2025 · 1 citation)
add structural features to the alignment objectiveinf-euroroad and ca-netscience, one-way noise · accuracyHelpedFUGAL: Feature-fortified Unrestricted Graph Alignment (NeurIPS 2024 · 1 citation)
select the top-b programs using individual upper bounds rather than greedy submodular optimizationAIDS · RMSEHurtGRAIL: Graph Edit Distance and Node Alignment using LLM-Generated Code (ICML 2025 · 0 citations)
add self-edgesMaximum retrieval · accuracyHelpedWhat Makes a Good Feedforward Computational Graph? (ICML 2025 · 0 citations)
Initialize P_g,l1* with P_MP or P_Loukas instead of dense P_raoCora, r=0.5 · RSA constantHurtTaxonomy of reduction matrices for Graph Coarsening (NeurIPS 2025 · 0 citations)
Replace the threshold (1+γ)·d_k with the more aggressive threshold d_1 + γ·d_kNavigable graphs · distance computations at fixed recallHurtDistance Adaptive Beam Search for Provably Accurate Graph-Based Nearest Neighbor Search (NeurIPS 2025 · 0 citations)

Baseline numbers that disagree

The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.

BaselineBenchmarkMetricReportedPapers
PCL-DCCiteseerAC0.468 – 0.6522 papers, e.g. AAAI 2018 and AAAI 2016

What reviewers will expect

Datasets and baselines named in the 10 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.

Datasets

  • MNIST · 2 of 10 papers
  • LINUX · 2 of 10 papers
  • Cora · 2 of 10 papers
  • Citeseer · 2 of 10 papers

Baselines

  • H2MN · 2 of 10 papers
  • ERIC · 2 of 10 papers
  • Greed · 2 of 10 papers
  • GEDGNN · 2 of 10 papers

Recent preprints

Loading recent arXiv preprints…

    Who publishes on it

    Institutions on the most papers of this approach since 2023 (of 42 papers; from OpenAlex and the papers' affiliations).

    Use it

    Download BibTeX of 39 key papersOpen the topic in the interactive map

    Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.

    Same keyword cluster

    Elsewhere in the topic

    Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).

    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