![]() ![]() Most of the existing HIN-oriented learning methods define a series of meta-paths. However, it is challenging to design a representation learning method for heterogeneous information networks (HINs) due to their diversity. ![]() Heterogeneous graph representation learning is to learn effective representations for nodes or (sub)graphs, which preserve node attributes and structural information. The results demonstrate that our framework achieves better performance than state-of-the-art graph embedding algorithms. ARWR-GE adopts an adversarial training scheme to enforce the latent codes to match a prior distribution, and by employing the skip-gram model, nodes in a random walk sequence are closer in the latent space.We evaluate our proposed framework by using three real-world datasets on link prediction, graph clustering, and visualization tasks. In this paper, we propose a novel graph embedding framework, Adversarial and Random Walk Regularized Graph Embedding (ARWR-GE), which jointly preserves structural and attribute information. #FLIXSTER CODES HOW TO#To mitigate this problem, we investigate how to enforce latent codes to match a prior distribution, and we introduce random walk to preserve high-order proximity in a graph. However, most existing graph convolutional network-based embedding algorithms not only ignore the data distribution of the latent codes but also lose the high-order proximity between nodes in a graph, leading to inferior embedding. Because of its powerful ability to model graph data, it is currently the best choice for graph embedding. The graph convolutional network is a neural network framework for machine learning on graphs. Graph embedding aims to represent node structural as well as attribute information into a low-dimensional vector space so that some downstream application tasks such as node classification, link prediction, community detection, and recommendation can be easily performed by using simple machine learning algorithms. The results show that our TGCNHF model can extract the spatio-temporal correlation from traffic data and the predictions overperform the state-of-art baselines on real-world traffic datasets. To test the proposed TGCNHF, a real-world travel time dataset collected in Beijing main urban area is used in comparison. Then, a fully connected layer with a SoftMax function accomplishes the NTR prediction. After stacking the length of time intervals to 1, a graph convolution is employed to extract the spatial correlation. In this model, features are divided into tendency-based and periodicity-based and handled respectively by two 1-D convolution layers on time axis. Further, this paper develops a temporal graph neural network with heterogeneous features (TGCNHF) to provide the real-time NTR. Aggregating the trips with the same origin in a specific time interval, we then introduce network travel risk (NTR) to evaluate the reliability of zone. In the trip-based reliability, OD pair rather than path or road is chosen as the object, which is different from the existing TTR. To estimate the trip-based reliability, this paper firstly defines the trip-based reliability as ‘the arriving late risk between an OD pair’. Compared with the TTR on a single road or path, the TTR of trip (delivery) seems more important for managers and logistics operators in decision making. The reliability of transport system is usually measured by travel time reliability (TTR). Green logistics and environmentally-friendly logistics necessitates transport system to be reliable for delivery. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |