Due to the rapid development of the sharing economy, shared bikes have become one of the most popular and convenient traveling tools in intelligent transport systems. Aiming to save the time spent on waiting for or searching bikes at bike stations, the operators of bike-sharing systems need to dynamically dispatch bikes. Predicting the number of bikes for each station can help to optimize the repository of bikes. The usage of bikes is affected by several uncertain factors, so bike number prediction becomes a challenging and difficult problem. To manage this problem, we propose an algorithm called bimodal Gaussian inhomogeneous Poisson (BGIP) to predict the number of bikes. The BGIP includes three steps. First, the inhomogeneous Poisson process is adopted to describe the process that people arrive at a bike station to pick up or return bikes. Second, the bimodal Gaussian function is used to describe the intensity function of inhomogeneous Poisson process. In order to dynamically uncover the changing trend in the usage state of bikes, we propose a method to measure the influences of external factors on the usage of bikes. Third, the number of bikes is predicted by calculating the mean usage of bikes on the basis of checking-out and checking-in sequences. Experiments demonstrated that our algorithm outperformed the baseline algorithms in solving the bike prediction problem: accurately predicting the number of bikes and determining whether there is at least one bike available at a bike station.

The problem of personalized next point-of-interest (POI) recommendation is significant and of practical value in location-based social networks (LBSNs). Due to the sparsity of data in regard to check-ins, POI recommendations remain a challenging problem. Previous work developed recommendation models by taking into consideration the properties of user's mobility, for example, spatio-temporal information and the similarity of movement rules. However, they ignore the influence of trip-purpose on user's mobility, and the memory effect of historical check-in behavior at the same location. To cope with this problem, a POI recommendation model with Spatio-Temporal effects based on Purpose Ranking (STPR) is proposed. STPR contains two important phases: (1) Purpose prediction: classifying the POIs in the spatio-temporal database into four categories. Each category corresponds to a purpose. Then, a purpose ranking model is constructed to model the selection of user's intended purpose for the trip. (2) The scoring of each candidate POI: the properties of spatial and temporal information are taken into consideration when calculating the score. For the spatial property, the kernel density estimation is used to estimate the visiting probability of a certain POI. For the temporal property, we focus on the time interval and the POI exploration mechanism. We find that POI exploration mechanism can reflect a user's mobility of visiting new POIs. Furthermore, a method based on Bayesian personalized ranking, called BPR, is proposed to estimate the time interval for unvisited POIs. Extensive experiments were conducted on two real datasets, and the experimental results demonstrate that the recommendation accuracy of the STPR model outperforms the state-of-the-art POI recommendation models with good runtime performance.

Accurately forecasting long-term future wind power is critical to achieve safe power grid integration. This problem is quite challenging due to wind power's high volatility and randomness. In this paper, we propose a novel time series forecasting method, namely Deep Conditional Generative Spatio-Temporal model (DCGST), and its high accuracy is achieved by tackling two critical issues simultaneously: a proper handling of the non-stationarity of multiple wind power time series, and a fine-grained modeling of their complicated yet dynamic spatio-temporal dependencies. Specifically, we first formally define the Spatio-Temporal Concept Drift (STCD) problem of wind power, and then we propose a novel deep conditional generative model to learn probabilistic distributions of future wind power values under STCD. Three different tailored neural networks are designed for distributions parameterization, including a graph-based prior network, an attention-based recognition network, and a stochastic seq2seq-based generation network. They are able to encode the dynamic spatio-temporal dependencies of multiple wind power time series and infer one-to-many mappings for future wind power generation. Compared to existing methods, DCGST can learn better spatio-temporal representations of wind power data and learn better uncertainties of data distribution to generate future values. Comprehensive experiments on real-world datasets including the largest public turbine-level wind power dataset verify the effectiveness, efficiency, generality and scalability of our method. Compared with state-of-the-art methods, DCGST can reduce the lone-term wind power prediction error by up to 22.8% and improve the online prediction efficiency by up to 76.8%.

Time series forecasting is a very vital research topic. The scale of time series in numerous industries has risen considerably in recent years as a result of the advancement of information technology. However, the existing algorithms pay little attention to generating large-scale time series. This article designs a state causality and adaptive covariance decomposition-based time series forecasting method (SCACD). As an observation sequence, the majority of time series is generated under the influence of hidden states. First, SCACD builds neural networks to adaptively estimate the mean and covariance matrix of latent variables; Then, SCACD employs causal convolution to forecast the distribution of future latent variables; Lastly, to avoid loss of information, SCACD applies a sampling approach based on Cholesky decomposition to generate latent variables and observation sequences. Compared to existing outstanding time series prediction models on six real datasets, the model can achieve long-term forecasting while also being lighter, and the forecasting accuracy is improved in the great majority of the prediction tasks.

Heterogeneous graph embedding has become a hot topic in network embedding in recent years and has been widely used in lots of practical scenarios. However, most of the existing heterogeneous graph embedding methods cannot make full use of all the auxiliary information. So we proposed a new method called Multi-Subgraph based Graph Convolution Network (MSGCN), which uses topology information, semantic information, and node feature information to learn node embedding vector. In MSGCN, the graph is firstly decomposed into multiple subgraphs according to the type of edges. Then convolution operation is adopted for each subgraph to obtain the node representations of each subgraph. Finally, the node representations are obtained by aggregating the representation vectors of nodes in each subgraph. Furthermore, we discussed the application of MSGCN with respect to a transductive learning task and inductive learning task, respectively. A node sampling method for inductive learning tasks to obtain representations of new nodes is proposed. This sampling method uses the attention mechanism to find important nodes and then assigns different weights to different nodes during aggregation. We conducted an experiment on three datasets. The experimental results indicate that our MSGCN outperforms the state-of-the-art methods in multi-class node classification tasks.