Graph convolutional neural networks exploit convolution operators, based on some neighborhood aggregating scheme, to compute representations of graphs. The most common operators only local topological information. To consider wider receptive fields, the mainstream approach is non-linearly stack multiple graph (GC) layers. In this way, however, interactions among GC parameters at different level...

Semi-supervised learning (SSL) provides a way to improve the performance of prediction models (e.g., classifier) via usage unlabeled samples. An effective and widely used method is construct graph that describes relationship between labeled Practical experience indicates quality significantly affects model performance. In this paper, we present visual analysis interactively constructs high-qual...

The decarbonization of the transport system requires a better understanding human mobility behavior to optimally plan and evaluate sustainable options (such as Mobility Service). Current analysis frameworks often rely on specific datasets or data-specific assumptions hence are difficult generalize other studies. In this work, we present workflow identify groups users with similar that appear ac...

We consider matrix completion for recommender systems from the point of view of link prediction on graphs. Interaction data such as movie ratings can be represented by a bipartite user-item graph with labeled edges denoting observed ratings. Building on recent progress in deep learning on graph-structured data, we propose a graph auto-encoder framework based on differentiable message passing on...

Although many graph-based clustering methods attempt to model the stationary diffusion state in their objectives, performance limits using a predefined graph. We argue that estimation of can be achieved by gradient descent over neural networks. specifically design Stationary Diffusion State Neural Estimation (SDSNE) exploit multiview structural graph information for co-supervised learning. expl...

Within this article, we present the application of the AutoSummENG method within the TAC 2010 AESOP challenge. We further present two novel evaluation methods based on n-gram graphs. The first method is called Merged Model Graph (MeMoG) and it uses the ngram graph framework to represent a set of documents with a single, “centroid” graph, offering state-of-the-art performance. The second method ...

The key to intelligent traffic control and guidance lies in accurate prediction of flow. Since flow data is nonlinear, complex, dynamic, order overcome these issues, graph neural network techniques are employed address challenges. For this reason, we propose a deep-learning architecture called AMGC-AT apply it real passenger dataset the Hangzhou metro for evaluation. Based on priori knowledge, ...