Alert button

Classifying Signals on Irregular Domains via Convolutional Cluster Pooling

Add code
Bookmark button
Alert button
Feb 13, 2019
Angelo Porrello, Davide Abati, Simone Calderara, Rita Cucchiara

Figure 1 for Classifying Signals on Irregular Domains via Convolutional Cluster Pooling
Figure 2 for Classifying Signals on Irregular Domains via Convolutional Cluster Pooling
Figure 3 for Classifying Signals on Irregular Domains via Convolutional Cluster Pooling
Figure 4 for Classifying Signals on Irregular Domains via Convolutional Cluster Pooling

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon

We present a novel and hierarchical approach for supervised classification of signals spanning over a fixed graph, reflecting shared properties of the dataset. To this end, we introduce a Convolutional Cluster Pooling layer exploiting a multi-scale clustering in order to highlight, at different resolutions, locally connected regions on the input graph. Our proposal generalises well-established neural models such as Convolutional Neural Networks (CNNs) on irregular and complex domains, by means of the exploitation of the weight sharing property in a graph-oriented architecture. In this work, such property is based on the centrality of each vertex within its soft-assigned cluster. Extensive experiments on NTU RGB+D, CIFAR-10 and 20NEWS demonstrate the effectiveness of the proposed technique in capturing both local and global patterns in graph-structured data out of different domains.

* 12 pages, 6 figures. To appear in the Proceedings of the 22nd International Conference on Artificial Intelligence and Statistics (AISTATS) 2019, Naha, Okinawa, Japan. PMLR: Volume 89  
View paper onarxiv icon

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon