Recent self-supervised pre-training methods on Heterogeneous Information Networks (HINs) have shown promising competitiveness over traditional semi-supervised Heterogeneous Graph Neural Networks (HGNNs). Unfortunately, their performance heavily depends on careful customization of various strategies for generating high-quality positive examples and negative examples, which notably limits their flexibility and generalization ability. In this work, we present SHGP, a novel Self-supervised Heterogeneous Graph Pre-training approach, which does not need to generate any positive examples or negative examples. It consists of two modules that share the same attention-aggregation scheme. In each iteration, the Att-LPA module produces pseudo-labels through structural clustering, which serve as the self-supervision signals to guide the Att-HGNN module to learn object embeddings and attention coefficients. The two modules can effectively utilize and enhance each other, promoting the model to learn discriminative embeddings. Extensive experiments on four real-world datasets demonstrate the superior effectiveness of SHGP against state-of-the-art unsupervised baselines and even semi-supervised baselines. We release our source code at: https://github.com/kepsail/SHGP.
Traffic signal control is a significant part of the construction of intelligent transportation. An efficient traffic signal control strategy can reduce traffic congestion, improve urban road traffic efficiency and facilitate people's lives. Existing reinforcement learning approaches for traffic signal control mainly focus on learning through a separate neural network. Such an independent neural network may fall into the local optimum of the training results. Worse more, the collected data can only be sampled once, so the data utilization rate is low. Therefore, we propose the Random Ensemble Double DQN Light (RELight) model. It can dynamically learn traffic signal control strategies through reinforcement learning and combine random ensemble learning to avoid falling into the local optimum to reach the optimal strategy. Moreover, we introduce the Update-To-Data (UTD) ratio to control the number of data reuses to improve the problem of low data utilization. In addition, we have conducted sufficient experiments on synthetic data and real-world data to prove that our proposed method can achieve better traffic signal control effects than the existing optimal methods.
Most existing Heterogeneous Information Network (HIN) embedding methods focus on static environments while neglecting the evolving characteristic of realworld networks. Although several dynamic embedding methods have been proposed, they are merely designed for homogeneous networks and cannot be directly applied in heterogeneous environment. To tackle above challenges, we propose a novel framework for incorporating temporal information into HIN embedding, denoted as Multi-View Dynamic HIN Embedding (MDHNE), which can efficiently preserve evolution patterns of implicit relationships from different views in updating node representations over time. We first transform HIN to a series of homogeneous networks corresponding to different views. Then our proposed MDHNE applies Recurrent Neural Network (RNN) to incorporate evolving pattern of complex network structure and semantic relationships between nodes into latent embedding spaces, and thus the node representations from multiple views can be learned and updated when HIN evolves over time. Moreover, we come up with an attention based fusion mechanism, which can automatically infer weights of latent representations corresponding to different views by minimizing the objective function specific for different mining tasks. Extensive experiments clearly demonstrate that our MDHNE model outperforms state-of-the-art baselines on three real-world dynamic datasets for different network mining tasks.
Graph Convolutional Network (GCN) has achieved extraordinary success in learning effective high-level representations of nodes in graphs. However, the study regarding Heterogeneous Information Network (HIN) is still limited, because the existing HIN-oriented GCN methods suffer from two deficiencies: (1) they cannot flexibly exploit all possible meta-paths, and some even require the user to specify useful ones; (2) they often need to first transform an HIN into meta-path based graphs by computing commuting matrices, which has a high time complexity, resulting in poor scalability. To address the above issues, we propose interpretable and efficient Heterogeneous Graph Convolutional Network (ie-HGCN) to learn representations of nodes in HINs. It automatically extracts useful meta-paths for each node from all possible meta-paths (within a length limit determined by the model depth), which brings good model interpretability. It directly takes the entire HIN as input and avoids intermediate HIN transformation. The carefully designed hierarchical aggregation architecture avoids computationally inefficient neighborhood attention. Thus, it is much more efficient than previous methods. We formally prove ie-HGCN evaluates the usefulness of all possible meta-paths within a length limit (model depth), show it intrinsically performs spectral graph convolution on HINs, and analyze the time complexity to verify its quasi-linear scalability. Extensive experimental results on three real-world networks demonstrate the superiority of ie-HGCN over state-of-the-art methods.