{ "id": "epc4a2yg2m", "slug": "hypergraph-learning", "title": "Hypergraph learning", "updated": "2026-10-15", "topic_path": [ { "id": "technology", "label": "Technology and the built world", "api_url": "https://www.edgepedia.org/api/v1/topics/technology" }, { "id": "technology.computing", "label": "Computing and digital systems", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing" }, { "id": "technology.computing.ai", "label": "Artificial intelligence and data", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai" }, { "id": "technology.computing.ai.ml_computation", "label": "Machine learning and neural computation", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.ml_computation" }, { "id": "technology.computing.ai.neural_deep", "label": "Neural networks and deep learning", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.neural_deep" }, { "id": "technology.computing.ai.nn_architectures", "label": "Neural network architectures", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.nn_architectures" }, { "id": "technology.computing.ai.nn_architectures.graph_networks", "label": "Graph neural network architectures", "api_url": "https://www.edgepedia.org/api/v1/topics/technology.computing.ai.nn_architectures.graph_networks" } ], "geo": [ { "id": "geo.nongeo.t2001.technology.computing.ai.neural_deep", "label": "Non-geographic · 2001 to 2020: Neural networks and deep learning", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai.neural_deep", "path": [ { "id": "geo.nongeo", "label": "Non-geographic", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo" }, { "id": "geo.nongeo.t2001", "label": "Non-geographic · 2001 to 2020", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001" }, { "id": "geo.nongeo.t2001.technology", "label": "Technology and the built world", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology" }, { "id": "geo.nongeo.t2001.technology.computing", "label": "Computing and digital systems", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing" }, { "id": "geo.nongeo.t2001.technology.computing.ai", "label": "Artificial intelligence and data", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai" }, { "id": "geo.nongeo.t2001.technology.computing.ai.ml_computation", "label": "Machine learning and neural computation", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai.ml_computation" }, { "id": "geo.nongeo.t2001.technology.computing.ai.neural_deep", "label": "Neural networks and deep learning", "api_url": "https://www.edgepedia.org/api/v1/geo/geo.nongeo.t2001.technology.computing.ai.neural_deep" } ] } ], "excerpt": "Hypergraph learning is machine learning on hypergraphs, whose hyperedges connect any number of nodes to model higher-order interactions like co-authorship, with deep-learning methods now predominant.", "snippet": "Hypergraph learning is machine learning on hypergraphs, whose hyperedges connect any number of nodes to model higher-order interactions like co-authorship, with deep-learning methods now predominant.", "node": "technology.computing.ai.nn_architectures.graph_networks", "markdown": "# Hypergraph learning\n\nHypergraph learning is machine learning on hypergraphs, structures whose hyperedges connect any number of nodes, so that a single relation can link three, a hundred, or thousands of vertices at once. A hyperedge models a higher-order interaction such as co-authorship, a group chat, a set of co-purchased items, or a protein interaction; forcing such interactions into a pairwise graph can cause information loss and measurable performance degradation. [1](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf) Learning on hypergraphs spans three families of techniques, spectral methods, proximity-preserving methods, and neural networks, with deep-learning approaches now predominant. [2](https://dl.acm.org/doi/10.1145/3605776)\n\n| Key fact | Value |\n|---|---|\n| Core spectral layer | \\( L = D_v^{-1/2} \\cdot H \\cdot B \\cdot D_e^{-1} \\cdot H^\\top \\cdot D_v^{-1/2} \\), with \\(H\\) the incidence matrix [3](https://www.dgl.ai/dgl_docs/en/1.0.x/notebooks/sparse/hgnn.html) |\n| Incidence matrix size | \\(O(mn)\\) for \\(n\\) nodes and \\(m\\) hyperedges [4](https://doi.org/10.48550/arxiv.2006.12278) |\n| HNHN convolution cost | \\(O(n\\delta_V d + nd^2 + md^2)\\), avoiding explicit \\(H\\) [4](https://doi.org/10.48550/arxiv.2006.12278) |\n| HGNN reported accuracy | 81.6% (Cora), 80.1% (Pubmed) over 100 runs [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235) |\n| HGNN under identical settings | 0.710 (Cora), 0.778 (Pubmed) in the HyFER benchmark [6](https://graph-learning-benchmarks.github.io/assets/papers/glb2021/Hyprgraph_Learning_Benchmark.pdf) |\n| UniGNN on DBLP | semi-supervised node classification raised from 77.4% to 88.8% [7](https://www.ijcai.org/proceedings/2021/0353.pdf) |\n| Applications | sequential, session-based, group, conversational, and point-of-interest recommendation; bioinformatics; time series; computer vision [8](https://dl.acm.org/doi/10.1145/3637528.3671457) |\n\n## How it works\n\nA weighted hypergraph \\(G=(V,E,w)\\) is described by an incidence matrix \\(H\\), with \\(h(v,e)=1\\) if node \\(v\\) belongs to hyperedge \\(e\\). Vertex degree is \\(d(v)=\\sum_e w(e)h(v,e)\\), and the adjacency is \\(A = HWH^\\top - D_v\\). [9](https://doi.org/10.7551/mitpress/7503.003.0205) Zhou, Huang, and Schölkopf generalized normalized-cut spectral clustering from graphs to hypergraphs: relaxing the hypergraph normalized cut yields a positive semidefinite hypergraph Laplacian whose eigendecomposition gives embedding and transductive (classifying only the given labeled-plus-unlabeled data) classification algorithms, with the classification function \\( f = (I - \\xi\\Theta)^{-1}y \\) for \\(\\xi\\in(0,1)\\). [9](https://doi.org/10.7551/mitpress/7503.003.0205) Their framework also has a random-walk interpretation. [10](https://dennyzhou.github.io/papers/hyper_tech.pdf)\n\nNeural hypergraph learning reuses this normalized operator as a convolution. The HGNN layer is \\( f(X^{(l)}, H; W^{(l)}) = \\sigma(L \\cdot X^{(l)} \\cdot W^{(l)}) \\) with \\( L = D_v^{-1/2} \\cdot H \\cdot B \\cdot D_e^{-1} \\cdot H^\\top \\cdot D_v^{-1/2} \\), where \\(B\\) is a diagonal hyperedge weight matrix. [3](https://www.dgl.ai/dgl_docs/en/1.0.x/notebooks/sparse/hgnn.html) Equivalently, each layer performs a two-stage node–hyperedge–node message pass: node features are transformed by a learned matrix, gathered into hyperedge representations via \\(H^\\top\\), and aggregated back to nodes via \\(H\\), with the degree matrices normalizing both stages. [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235)\n\n## How it is done\n\nA practitioner first builds the hypergraph. In HGNN's visual object construction, each hyperedge connects one vertex and its \\(K\\) nearest neighbors by [Euclidean distance](https://www.edgepedia.org/euclidean-distance), giving \\(N\\) hyperedges over \\(K+1\\) vertices and \\(H \\in \\mathbb{R}^{N \\times N}\\); in citation networks, each hyperedge links a vertex and its graph neighbors. [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235) [Hypergraph](https://www.edgepedia.org/hypergraph) expression can also use clique expansion, adaptive expansion, star expansion, or line expansion. [1](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf)\n\nSecond, the incidence matrix and degree matrices are computed: node degree counts the hyperedges containing a node, hyperedge degree counts the nodes in a hyperedge, given by \\(D_V=\\operatorname{diag}(H w)\\) and \\(D_E=\\operatorname{diag}(H^\\top\\mathbf{1})\\); plain sums of \\(H\\) apply in the unweighted case. [3](https://www.dgl.ai/dgl_docs/en/1.0.x/notebooks/sparse/hgnn.html) Third, propagation layers are stacked. HGNN approximates the spectral convolution with truncated [Chebyshev polynomials](https://www.edgepedia.org/chebyshev-polynomials) to avoid the \\(O(n^2)\\) cost of forward and inverse hypergraph Fourier transforms; with \\(K=1\\) and \\(\\lambda_{\\max}\\approx 2\\) the filter simplifies to \\( g \\star x \\approx \\theta_0 x - \\theta_1 D_v^{-1/2} \\cdot H \\cdot W \\cdot D_e^{-1} \\cdot H^\\top \\cdot D_v^{-1/2} x \\). [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235) Training then targets node classification, hyperedge classification or prediction, clustering, or recommendation. HNHN's convolution costs \\(O(n\\delta_V d + nd^2 + md^2)\\) for \\(n\\) nodes, \\(m\\) hyperedges, average vertex degree \\(\\delta_V\\), and hidden dimension \\(d\\), and avoids instantiating the incidence matrix, which has size \\(O(mn)\\). [4](https://doi.org/10.48550/arxiv.2006.12278)\n\n## Origin\n\nAn influential early framework for hypergraph clustering, classification, and embedding was presented by Dengyong Zhou, Jiayuan Huang, and [Bernhard Schölkopf](https://www.edgepedia.org/bernhard-scholkopf), who generalized spectral clustering to hypergraphs and derived hypergraph embedding and transductive classification algorithms from it, appearing at NeurIPS in 2006 and published in 2007 in The MIT Press eBooks. [9](https://doi.org/10.7551/mitpress/7503.003.0205) Earlier still, the first discussion of vector representation learning for hypergraph nodes and hyperedges dates to the 1980s, aimed at 2D visualization. [2](https://dl.acm.org/doi/10.1145/3605776) On UCI datasets such as zoo (101 animals, 16 predictor attributes including 15 Boolean and one numeric, 7 classes), the hypergraph method was consistently better than the pairwise baseline. [10](https://dennyzhou.github.io/papers/hyper_tech.pdf)\n\nThe neural turn came from Yifan Feng and colleagues, whose HGNN, posted as an arXiv preprint in 2018 [11](https://doi.org/10.48550/arxiv.1809.09401) and published at AAAI 2019, presented the first [hypergraph neural network](https://www.edgepedia.org/hypergraph-neural-network) with a designed hyperedge convolution operation. [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235)\n\n## Variants\n\nNamed variants differ mainly in how the two-stage message pass is normalized and parameterized. **HGNN+** (Yue Gao, Yifan Feng, Shuyi Ji, and Rongrong Ji, IEEE TPAMI 2022) extends HGNN into a general framework for multi-modal, multi-type correlations using hyperedge groups and an adaptive fusion strategy, with convolution performed in the spatial domain. [12](https://doi.org/10.1109/tpami.2022.3182052) **HyperGCN** (Naganand Yadati and colleagues, arXiv 2018) trains graph convolutional networks on hypergraphs. [13](https://doi.org/10.48550/arxiv.1809.02589) **HNHN** (Yihe Dong, Will Sawin, and [Yoshua Bengio](https://www.edgepedia.org/yoshua-bengio), 2020) applies nonlinear activations to both hypernodes and hyperedges, with node update \\( X'_V = \\sigma(D_V^{-\\beta} H \\cdot X'_E \\cdot W_V + b_V) \\) and hyperedge update \\( X'_E = \\sigma(D_E^{-\\alpha} H^\\top \\cdot X_V \\cdot W_E + b_E) \\), plus normalization hyperparameters \\(\\alpha\\) and \\(\\beta\\) that adjust the weight of high-cardinality hyperedges and high-degree vertices. [4](https://doi.org/10.48550/arxiv.2006.12278) **UniGNN** (Jing Huang and Jie Yang, 2021) generalizes GCN, GAT, GIN, and [GraphSAGE](https://www.edgepedia.org/graphsage) to hypergraphs through permutation-invariant functions \\(\\varphi_1\\) and \\(\\varphi_2\\), reducing to standard GNNs on ordinary graphs; its UniGCNII variant adds initial residuals and identity mappings against oversmoothing. [7](https://www.ijcai.org/proceedings/2021/0353.pdf) • [14](https://doi.org/10.48550/arxiv.2105.00956) **AllSet** (Eli Chien, Chao Pan, Jianhao Peng, and Olgica Milenkovic, 2021) frames both stages as multiset functions and generalizes most HNNs, including HGNN, HNHN, HCHA, HyperSAGE, and HyperGCN. [15](https://doi.org/10.48550/arxiv.2106.13264) • [16](https://arxiv.org/html/2310.07684) **HCHA** (Song Bai, Feihu Zhang, and Philip H.S. Torr, 2020) adds hypergraph attention. [17](https://doi.org/10.1016/j.patcog.2020.107637) A separate line learns hyperedge-dependent node embeddings: the HNN framework of Ryan Aponte and colleagues (2022) jointly learns hyperedge and hyperedge-dependent node embeddings. [18](https://doi.org/10.48550/arxiv.2212.14077)\n\n## Applications\n\nHypergraph neural networks have been applied to sequential, session-based, group, conversational, and point-of-interest recommendation, where a session's clicked or purchased items are connected by a hyperedge. [1](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf) • [8](https://dl.acm.org/doi/10.1145/3637528.3671457) Survey coverage also includes bioinformatics and medical science, time series analysis, and computer vision. [8](https://dl.acm.org/doi/10.1145/3637528.3671457) In visual object recognition, HGNN gains 8.3%, 10.4%, and 8.1% over GCN on the NTU dataset when GVCNN, MVCNN, and GVCNN+MVCNN features are used. [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235)\n\n## Limitations and alternatives\n\n**Clique expansion loses information.** HGNN and HCHA without attention are mathematically equivalent to ordinary graph convolution on the clique expansion, which does not use all structural information of the hypergraph. [4](https://doi.org/10.48550/arxiv.2006.12278) No bijective transformation exists between a hypergraph and its clique expansion, and clique or star expansions fail to model hyperedge–vertex dependency and interactions between hyperedges. [19](https://ar5iv.labs.arxiv.org/html/2101.07773) Non-reductive alternatives include star expansion, line expansion, and tensor representation. [8](https://dl.acm.org/doi/10.1145/3637528.3671457)\n\n**Oversmoothing and depth.** Clique-expansion-based HNNs often perform best shallow; deeper layers cause oversmoothing. [16](https://arxiv.org/html/2310.07684) UniGCNII addresses this with residual and identity connections. [7](https://www.ijcai.org/proceedings/2021/0353.pdf)\n\n**Scalability and construction.** Spectral methods scale poorly with \\(|V|\\) because the eigenproblem's matrices must be stored in memory. [2](https://dl.acm.org/doi/10.1145/3605776) HGNN's open-source implementation runs out of memory on Pubmed and DBLP. [6](https://graph-learning-benchmarks.github.io/assets/papers/glb2021/Hyprgraph_Learning_Benchmark.pdf) Pipelines struggle when hyperedges are large; on 20Newsgroups, hyperedges have a median of 537 nodes. [16](https://arxiv.org/html/2310.07684) There is no standardized way to construct hypergraphs, since node and hyperedge definitions vary across applications, and scalability on large datasets is not fully validated. [20](https://link.springer.com/content/pdf/10.1007/s40305-025-00630-y.pdf) HNN's authors note that most methods assume nodes in the same hyperedge should be represented similarly, which fails on noisy, partially observed hyperedges of arbitrary size. [21](https://ar5iv.labs.arxiv.org/html/2212.14077) Naively extending GNN techniques to hypergraphs often yields suboptimal performance, so tailored designs remain necessary. [1](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf)\n\nAgainst alternatives, HNNs have delivered substantial improvements over GNNs in node classification, edge prediction, and clustering, including state-of-the-art results on DBLP, Trivago, and House leaderboards. [1](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf) Higher-order learning also proceeds through motifs in dyadic graphs and through simplicial complexes, with hypergraphs one of the two direct representations. [22](https://par.nsf.gov/biblio/10528348-higher-order-networks-representation-learning-survey) A first systematic study of liftings found that the choice of lifting often matters more than the choice of HNN architecture, and no single lifting dominates across datasets. [23](https://proceedings.mlr.press/v326/montagna26a.html) Reported numbers depend heavily on settings: HGNN's paper reports 81.6% on Cora and 80.1% on Pubmed over 100 runs, [5](https://ojs.aaai.org/index.php/AAAI/article/view/4235) but under HyFER's identical settings HGNN reaches 0.710 (Cora) and 0.778 (Pubmed), with HNHN at 0.660 and 0.771 and HAT at 0.607 and 0.764. [6](https://graph-learning-benchmarks.github.io/assets/papers/glb2021/Hyprgraph_Learning_Benchmark.pdf)\n\n## References\n\n1. [A Tutorial on Hypergraph Neural Networks: An In-Depth and Step-By-Step Guide (CIKM 2025)](http://dmlab.kaist.ac.kr/~kijungs/papers/hnnCIKM2025.pdf)\n2. [A Survey on Hypergraph Representation Learning (ACM Computing Surveys)](https://dl.acm.org/doi/10.1145/3605776)\n3. [Hypergraph Neural Networks, DGL documentation tutorial](https://www.dgl.ai/dgl_docs/en/1.0.x/notebooks/sparse/hgnn.html)\n4. [Dong, Yihe, Sawin, Will, Bengio, Yoshua (2020). HNHN: Hypergraph Networks with Hyperedge Neurons. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.2006.12278)\n5. [Hypergraph Neural Networks (HGNN, AAAI 2019; facts and excerpts merged from the arXiv preprint arXiv:1809.09401 of the same paper)](https://ojs.aaai.org/index.php/AAAI/article/view/4235)\n6. [HyFER: A Framework for Making Hypergraph Learning Easy, Scalable and Benchmarkable](https://graph-learning-benchmarks.github.io/assets/papers/glb2021/Hyprgraph_Learning_Benchmark.pdf)\n7. [UniGNN: a Unified Framework for Graph and Hypergraph Neural Networks (IJCAI 2021)](https://www.ijcai.org/proceedings/2021/0353.pdf)\n8. [A Survey on Hypergraph Neural Networks: An In-Depth and Step-By-Step Guide (KDD 2024; merged with its arXiv version arXiv:2404.01039)](https://dl.acm.org/doi/10.1145/3637528.3671457)\n9. [Dengyong Zhou, Jiayuan Huang, Bernhard Schölkopf (2007). Learning with Hypergraphs: Clustering, Classification, and Embedding. The MIT Press eBooks.](https://doi.org/10.7551/mitpress/7503.003.0205)\n10. [Beyond Pairwise Classification and Clustering Using Hypergraphs (Zhou et al. technical report)](https://dennyzhou.github.io/papers/hyper_tech.pdf)\n11. [Feng, Yifan and colleagues (2018). Hypergraph Neural Networks. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.1809.09401)\n12. [Yue Gao and colleagues (2022). HGNN + : General Hypergraph Neural Networks. IEEE Transactions on Pattern Analysis and Machine Intelligence.](https://doi.org/10.1109/tpami.2022.3182052)\n13. [Yadati, Naganand and colleagues (2018). HyperGCN: A New Method of Training Graph Convolutional Networks on Hypergraphs. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.1809.02589)\n14. [Huang, Jing, Yang, Jie (2021). UniGNN: a Unified Framework for Graph and Hypergraph Neural Networks. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.2105.00956)\n15. [Chien, Eli and colleagues (2021). You are AllSet: A Multiset Function Framework for Hypergraph Neural Networks. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.2106.13264)\n16. [Hypergraph Neural Networks through the Lens of Message Passing (MultiSet)](https://arxiv.org/html/2310.07684)\n17. [Song Bai, Feihu Zhang, Philip H.S. Torr (2020). Hypergraph convolution and hypergraph attention. Pattern Recognition.](https://doi.org/10.1016/j.patcog.2020.107637)\n18. [Aponte, Ryan and colleagues (2022). A Hypergraph Neural Network Framework for Learning Hyperedge-Dependent Node Embeddings. arXiv (Cornell University).](https://doi.org/10.48550/arxiv.2212.14077)\n19. [Learning over Families of Sets - Hypergraph Representation Learning for Higher Order Tasks](https://ar5iv.labs.arxiv.org/html/2101.07773)\n20. [Recent Advances in Hypergraph Neural Networks (Springer, 2025; merged with its arXiv version arXiv:2503.07959)](https://link.springer.com/content/pdf/10.1007/s40305-025-00630-y.pdf)\n21. [A Hypergraph Neural Network Framework for Learning Hyperedge-Dependent Node Embeddings (HNN)](https://ar5iv.labs.arxiv.org/html/2212.14077)\n22. [Higher-Order Networks Representation and Learning: A Survey (KDD Explorations, 2024)](https://par.nsf.gov/biblio/10528348-higher-order-networks-representation-learning-survey)\n23. [Lift me up: the impact of liftings on hypergraph neural networks (PMLR)](https://proceedings.mlr.press/v326/montagna26a.html)\n\n---\n*Topic: Encyclopedia › Technology and the built world › Computing and digital systems › Artificial intelligence and data › Machine learning and neural computation › Neural networks and deep learning › Neural network architectures › Graph neural network architectures*\n\n*Initially written Sep 29, 2026 · Edited: Oct 15, 2026*\n\n*Copyright 2026 EdgeChat AI, a subsidiary of Biostate AI.*\n\nLicense: Edgepedia Community License 1.0, https://www.edgepedia.org/edgepedia/license\n", "same_as": [], "url": "https://www.edgepedia.org/hypergraph-learning", "markdown_url": "https://www.edgepedia.org/hypergraph-learning.md", "license": { "name": "Edgepedia Community License 1.0", "url": "https://www.edgepedia.org/edgepedia/license", "summary": "Free with credit, commercial use included. AI training is open to everyone. For other uses, organizations over USD 100M in revenue or 100M monthly users license separately.", "spdx": "LicenseRef-Edgepedia-Community-1.0" }, "credit": "\"Hypergraph learning\", Edgepedia (EdgeChat), https://www.edgepedia.org/hypergraph-learning. Edgepedia Community License 1.0.", "credit_md": "\"[Hypergraph learning](https://www.edgepedia.org/hypergraph-learning)\", Edgepedia (EdgeChat), [https://www.edgepedia.org/hypergraph-learning](https://www.edgepedia.org/hypergraph-learning). [Edgepedia Community License 1.0](https://www.edgepedia.org/edgepedia/license).", "credit_html": "\"Hypergraph learning\", Edgepedia (EdgeChat), https://www.edgepedia.org/hypergraph-learning. Edgepedia Community License 1.0.", "speakable": "Hypergraph learning is machine learning on hypergraphs, whose hyperedges connect any number of nodes to model higher-order interactions like co-authorship, with deep-learning methods now predominant." }