Deep learning on graphs has attracted significant interests recently. However, most of the works have focused (semi-) supervised learning, resulting in shortcomings including heavy label reliance, poor generalization, and weak robustness. To address these issues, self-supervised (SSL), which extracts informative knowledge through well-designed pretext tasks without relying manual labels, become a promising trending paradigm for graph data. Different from SSL other domains like computer...

Considering the ill-posed nature, contrastive regularization has been developed for single image dehazing, introducing information from negative images as a lower bound. However, samples are non-consensual, negatives usually represented distantly clear (i.e., positive) image, leaving solution space still under-constricted. Moreover, interpretability of deep dehazing models is underexplored towards physics hazing process. In this paper, we propose novel curricular targeted at consensual...

In recent years, graph neural networks (GNNs) have emerged as a successful tool in variety of graph-related applications. However, the performance GNNs can be deteriorated when noisy connections occur original structures; besides, dependence on explicit structures prevents from being applied to general unstructured scenarios. To address these issues, recently deep structure learning (GSL) methods propose jointly optimize along with GNN under supervision node classification task. Nonetheless,...

Robust prediction of citywide traffic flows at different time periods plays a crucial role in intelligent transportation systems. While previous work has made great efforts to model spatio-temporal correlations, existing methods still suffer from two key limitations: i) Most models collectively predict all regions' without accounting for spatial heterogeneity, i.e., regions may have skewed flow distributions. ii) These fail capture the temporal heterogeneity induced by time-varying patterns,...

Multivariate time series forecasting has long received significant attention in real-world applications, such as energy consumption and traffic prediction. While recent methods demonstrate good abilities, they have three fundamental limitations. (i). <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Discrete neural architectures:</i> Interlacing individually parameterized spatial temporal blocks to encode rich underlying patterns leads...

This study utilized the hierarchical linear model and trust as a mediator; designated leader emotional intelligence team-level dimension; transformational leadership, transactional in supervisor individual-level dimensions. An analysis was performed to explore relationships between these individual- variables job performance of real estate brokers, well potential role trust-oriented leadership mediator for intelligence, individual performance. The empirical results revealed that had direct,...

The real-time discovery of local events (e.g., protests, crimes, disasters) is great importance to various applications, such as crime monitoring, disaster alarming, and activity recommendation. While this task was nearly impossible years ago due the lack timely reliable data sources, recent explosive growth in geo-tagged tweet brings new opportunities it. That said, how extract quality from streams real time remains largely unsolved so far.

Anomaly detection from graph data has drawn much attention due to its practical significance in many critical applications including cybersecurity, finance, and social networks. Existing mining machine learning methods are either shallow that could not effectively capture the complex interdependency of or autoencoder fully exploit contextual information as supervision signals for effective anomaly detection. To overcome these challenges, this paper, we propose a novel method, Self-Supervised...

Despite the rapid advance of unsupervised anomaly detection, existing methods require to train separate models for different objects. In this work, we present UniAD that accomplishes detection multiple classes with a unified framework. Under such challenging setting, popular reconstruction networks may fall into an "identical shortcut", where both normal and anomalous samples can be well recovered, hence fail spot outliers. To tackle obstacle, make three improvements. First, revisit...

Anomaly detection on graphs plays a significant role in various domains, including cybersecurity, e-commerce, and financial fraud detection. However, existing methods graph anomaly usually consider the view single scale of graphs, which results their limited capability to capture anomalous patterns from different perspectives. Towards this end, we introduce novel framework, namely ANEMONE, simultaneously identify anomalies multiple scales. Concretely, ANEMONE first leverages neural network...

With the continuous development of deep learning in field image generation models, a large number vivid forged faces have been generated and spread on Internet. These high-authenticity artifacts could grow into threat to society security. Existing face forgery detection methods directly utilize obtained public shared or centralized data for training but ignore personal privacy security issues when couldn't be centralizedly real-world scenarios. Additionally, different distributions caused by...

CircRNA has been shown to be involved in the occurrence of many diseases. Several computational frameworks have proposed identify circRNA-disease associations. Despite existing methods obtained considerable successes, these still require improved as their performance may degrade due sparsity data and problem memory overflow. We develop a novel framework called LGCDA predict associations by fusing local global features solve above mentioned problems. First, we construct closed subgraphs using...

The rapid growth of scientific papers makes it difficult to find relevant and appropriate citations. Context-aware citation recommendation aims overcome this problem by providing a list given short passage text. In paper, we propose long-short-term memory (LSTM)based model for context-aware recommendation, which first learns the distributed representations contexts separately based on LSTM, then measures relevance learned representation papers. Finally, with high scores are selected as list....

Unsupervised spatial representation learning aims to automatically identify effective features of geographic entities (i.e., regions) from unlabeled yet structural geographical data. Existing network embedding methods can partially address the problem by: (1) regarding a region as node in order reformulate into embedding; (2) graph embedding. However, these studies be improved by preserving intra-region structures, which are represented multiple graphs, leading reformulation collective...

Graph contrastive learning (GCL) alleviates the heavy reliance on label information for graph representation (GRL) via self-supervised schemes. The core idea is to learn by maximising mutual similar instances, which requires similarity computation between two node instances. However, GCL inefficient in both time and memory consumption. In addition, normally a large number of training epochs be well-trained large-scale datasets. Inspired an observation technical defect (i.e., inappropriate...

Multivariate time-series anomaly detection is critically important in many applications, including retail, transportation, power grid, and water treatment plants. Existing approaches for this problem mostly employ either statistical models which cannot capture the nonlinear relations well or conventional deep learning (DL) e.g., convolutional neural network (CNN) long short-term memory (LSTM) that do not explicitly learn pairwise correlations among variables. To overcome these limitations,...

Clustering short texts (such as news titles) by their meaning is a challenging task. The semantic hashing approach encodes the of text into compact binary code. Thus, to tell if two have similar meanings, we only need check they codes. encoding created deep neural network, which trained on represented word-count vectors (bag-of-word representation). Unfortunately, for such search queries, tweets, or titles, representations are insufficient capture underlying semantics. To cluster propose add...

Since the Chinese national carbon trading market was launched in 2017, price has become an important research topic. This study constructs ‘China index’, and then a SVAR model with China index, EU industrial Securities Index energy index (CSI), air quality (AQI) HS300 to prices China. The result shows that AQI have direct effect on price. Meanwhile, CSI indirect price, is slightly positive. In addition, volatility of China’s mainly internally driven, while other economic variables examined...

This study examines the effects of leader emotional intelligence, leadership styles (transformational and transactional), organizational commitment, trust on job performance. A questionnaire was administered to participants, who were real estate brokers in Kaohsiung City. Of 980 questionnaires administered, 348 valid responses received, indicating an effective response rate 35.5%. Structural equation modeling used for analysis. The results show that intelligence has a significant positive...

With the increasing number of scientific papers, researchers find it more and difficult to obtain relevant appropriate papers cite. Citation recommendation aims overcome this problem by providing a reference paper list for given manuscript. In paper, we propose bibliographic network representation (BNR) model, which simultaneously incorporates structure content different kinds objects (authors, venues) efficient recommendation. The proposed model also makes personalized citation possible, is...

MgO@SiO 2 nanocapsules were synthesized, and the silica coating prevents rapid degradation of nanocrystal controlled magnesium release, which helps to alleviate cartilage damage inflammation caused by OA.