“…G RAPH data processing using neural networks has been broadly attracting more and more research interests recently. Graph convolutional networks (GCNs) [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39] are a family of graphbased neural networks that extend convolutional neural networks (CNNs) to extract local features in general graphs with irregular input structures. The irregularity of a graph, including the orderless nodes and connections, however, makes the GCNs difficult to design as well as training from local patterns.…”