Timezone: »

Haar Graph Pooling
Yuguang Wang · Ming Li · Zheng Ma · Guido Montufar · Xiaosheng Zhuang · Yanan Fan

Wed Jul 15 03:00 PM -- 03:45 PM & Thu Jul 16 02:00 AM -- 02:45 AM (PDT) @ Virtual #None

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 --- \emph{HaarPooling}. HaarPooling implements a cascade of pooling operations; it is computed by following a sequence of clusterings of the input graph. A HaarPooling layer transforms a given input graph to an output graph with a smaller node number and the same feature dimension; the compressive Haar transform filters out fine detail information in the Haar wavelet domain. In this way, all the HaarPooling layers together synthesize the features of any given input graph into a feature vector of uniform size. Such transforms provide a sparse characterization of the data and preserve the structure information of the input graph. GNNs implemented with standard graph convolution layers and HaarPooling layers achieve state of the art performance on diverse graph classification and regression problems.

Author Information

Yu Guang Wang (UNSW; MPI MIS)
Ming Li (Zhejiang Normal University)

I received my PhD degree from the Department of Computer Science and IT from La Trobe University, Australia. Currently, I am a "Shuang Long Scholar" Distinguished Professor with the Department of Educational Technology, Zhejiang Normal University, China. Before that, I completed two Postdoctoral Fellowship positions with the Department of Mathematics and Statistics, La Trobe University, Australia, and the Department of Information Technology in Education, South China Normal University, China, respectively. My research outputs appear in top-tier journals and conferences, including Neural Networks, Information Sciences, IEEE Transactions on Cybernetics (one paper is ranked as ESI Highly Cited Paper ), Applied Mathematical Modelling, Applied Mathematics and Computation, ICML, IJCNN. I am acting as a regular reviewer for top journals including IEEE TNNLS, IEEE TCYB, IEEE TKDE, Neural Networks, Information Sciences, Neurocomputing, Applied Soft Computing, Applied Mathematical Modelling, WIREs Data Mining and Knowledge Discovery, etc. I was the recipient of outstanding reviewer of IEEE TCYB in 2016 and 2017.

Zheng Ma (Princeton University)
Guido Montufar (UCLA Math / Stat; MPI MIS)
Xiaosheng Zhuang (City University of Hong Kong)
Yanan Fan (University of New South Wales)