# Hierarchical Graph Capsule Network

arXiv: https://arxiv.org/abs/2012.08734 · HTML: https://mlsanity.com/paper/2012.08734/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Dec 2020.

## Built on by (most cited first, 11 of 11)

- [Graph Pooling for Graph Neural Networks: Progress, Challenges, and Opportunities](https://mlsanity.com/paper/2204.07321/) (2022)
- [Distribution Knowledge Embedding for Graph Pooling](https://mlsanity.com/paper/2109.14333/) (2021)
- [Dynamic Hypergraph Structure Learning for Traffic Flow Forecasting](https://mlsanity.com/paper/2309.12028/) (2023)
- [Visual Dependency Transformers: Dependency Tree Emerges from Reversed Attention](https://mlsanity.com/paper/2304.03282/) (2023)
- [Topological Pooling on Graphs](https://mlsanity.com/paper/2303.14543/) (2023)
- [Hierarchical Capsule Prediction Network for Marketing Campaigns Effect](https://mlsanity.com/paper/2208.10113/) (2022)
- [Incorporating Heterophily into Graph Neural Networks for Graph Classification](https://mlsanity.com/paper/2203.07678/) (2022)
- [Capsule Network Projectors are Equivariant and Invariant Learners](https://mlsanity.com/paper/2405.14386/) (2024)
- [Relaxing Continuous Constraints of Equivariant Graph Neural Networks for Physical Dynamics Learning](https://mlsanity.com/paper/2406.16295/) (2024)
- [GPN: A Joint Structural Learning Framework for Graph Neural Networks](https://mlsanity.com/paper/2205.05964/) (2022)
- [Reducing Smoothness with Expressive Memory Enhanced Hierarchical Graph Neural Networks](https://mlsanity.com/paper/2504.00349/) (2025)

## Abstract (arXiv)

Graph Neural Networks (GNNs) draw their strength from explicitly modeling the topological information of structured data. However, existing GNNs suffer from limited capability in capturing the hierarchical graph representation which plays an important role in graph classification. In this paper, we innovatively propose hierarchical graph capsule network (HGCN) that can jointly learn node embeddings and extract graph hierarchies. Specifically, disentangled graph capsules are established by identifying heterogeneous factors underlying each node, such that their instantiation parameters represent different properties of the same entity. To learn the hierarchical representation, HGCN characterizes the part-whole relationship between lower-level capsules (part) and higher-level capsules (whole) by explicitly considering the structure information among the parts. Experimental studies demonstrate the effectiveness of HGCN and the contribution of each component.