In this paper, we study the efficiency of Graph Transformer Network for noisy label propagation in task classifying video anomaly actions. Given a weak supervised dataset, our methods focus on improving quality generated labels and use training classifier with deep network. From full-length video, properties each segmented can be decided through their relationship other video. Therefore, employ...

Applications in various domains rely on processing graph streams, e.g., communication logs of a cloud-troubleshooting system, roadnetwork traffic updates, and interactions on a social network. A labeled-graph stream refers to a sequence of streamed edges that form a labeled graph. Label-aware applications need to filter the graph stream before performing a graph operation. Due to the large volu...

Many web-based application areas must infer label distributions starting from a small set of sparse, noisy labels. Previous work has shown that graph-based propagation can be very effective at finding the best label distribution across nodes, starting from partial information and a weightedconnection graph. In their work on video recommendations, Baluja et al. showed high-quality results using ...

A covering projection from a graph G onto a graph H is a \local isomorphism": a mapping from the vertex set of G onto the vertex set of H such that, for every v 2 V (G), the neighborhood of v is mapped bijectively onto the neighborhood (in H) of the image of v. We investigate two concepts that concern graph covers of regular graphs. The rst one is called \multicovers": we show that for any regu...

We consider the problem of periodic graph traversal [2], which has previously been studied in a variety of settings [2, 5, 3, 7, 13, 12, 10, 11]. The problem of periodic graph traversal is concerned with an agent having to visit every node of a graph and return to its start location and state. The periodic graph traversal problem can be extended by labeling each node to help the agent explore t...

Multi-view partial multi-label learning (MVPML) is a fundenmental problem where each sample linked to multiple kinds of features and candidate labels, including ground-truth noise labels. The key MVPML how manipulate the recover labels from label set. To this end, study designs novel Graph-based Partial Multi-label model named as GMPM, which combines multi-view information detection, valuable s...