Haar Graph Pooling

doi:doi:10.57702/1wsog130

Haar Graph Pooling

Deep Graph Neural Networks (GNNs) are useful models for graph classification and graph-based regression tasks. In these tasks, graph pooling is a critical ingredient by which GNNs adapt to input graphs of varying size and structure. We propose a new graph pooling operation based on compressive Haar transforms — HaarPooling. HaarPooling implements a cascade of pooling operations; it is computed by following a sequence of clusterings of the input graph.

BibTex:

@dataset{Yu_Guang_Wang_and_Ming_Li_and_Zheng_Ma_and_Guido_Mont´ufar_and_Xiaosheng_Zhuang_and_Yanan_Fan_2024,
    abstract = {Deep Graph Neural Networks (GNNs) are useful models for graph classification and graph-based regression tasks. In these tasks, graph pooling is a critical ingredient by which GNNs adapt to input graphs of varying size and structure. We propose a new graph pooling operation based on compressive Haar transforms — HaarPooling. HaarPooling implements a cascade of pooling operations; it is computed by following a sequence of clusterings of the input graph.},
    author = {Yu Guang Wang and Ming Li and Zheng Ma and Guido Mont´ufar and Xiaosheng Zhuang and Yanan Fan},
    doi = {10.57702/1wsog130},
    institution = {No Organization},
    keyword = {'Deep Learning', 'Graph Neural Networks', 'Graph Pooling', 'HaarPooling'},
    month = {dec},
    publisher = {TIB},
    title = {Haar Graph Pooling},
    url = {https://service.tib.eu/ldmservice/dataset/haar-graph-pooling},
    year = {2024}
}