Search

The complex and varied urban road conditions have always been a difficult and a pivotal component in the study of driverless technology especially at intersections. In China, it is necessary for driverless cars to understand the gestures of traffic police. To identify the traffic police gesture at the intersection, the key step is to detect the traffic police at the intersection. At present, the research on traffic police detection is still in its infancy, there exists common problems such as slow detection speed and other real time problems in this method, and there is not a standardized traffic police data set ether. For the real time problems, this paper introduces the attention mechanism, and proposes a new real-time detection method of traffic police based on attention mechanism. The method proposed in this paper has strong robustness and can quickly complete the target detection task. For the data set problem, this paper analyzes and discloses similar data sets published through research. In the meanwhile, this paper collects 24,530 video data on the actual road, and it extracts and saves 12,000 pictures containing traffic police at frame rate of 30FPS as traffic police detection training and validated data set. [ABSTRACT FROM AUTHOR]

Traffic flow prediction is an important component for self-driving. Traffic flow is closely related to population distribution, and the traffic flow is not only related to the absolute number of human population but also to their concerns and interests. Accurate spatio-temporal web traffic flow prediction is critical in many applications, such as bandwidth allocation, anomaly detection, congestion control and admission control. Most existing traffic flow prediction methods use models based on time-series analysis and remain inadequate for many real-world applications. Web traffic flow is found to be strongly associated with the spatio-temporal distribution of the population. Increasingly, it is critical to understand and make decisions based on the relationship between population patterns and web traffic flow patterns. It has been proven that different people have different responses to web events. Due to the complexity of spatial data structures and the huge volume of web traffic flow log data, it is difficult to routinely find the relationship between web events and population distributions without an appropriate processing framework. In this paper, we propose an innovative framework named GeoTrafficPredict to support the accurate spatio-temporal prediction of web traffic flow. GeoTrafficPredict provides a machine learning platform to learn the spatio-temporal pattern of traffic flow and use the pattern to predict the trend in both spatial and temporal dimension. Also, GeoTrafficPredict provide data aggregation portal and cloud-based computation function. GeoTrafficPredict deploys a series of computational images in a cloud computing environment, and the implementation on China's CSTNET illustrates the performance of our platform. [ABSTRACT FROM AUTHOR]

Modeling interpretable artificial intelligence (AI) for flood forecasting represents a serious challenge: both accuracy and interpretability are indispensable. Because of the uncertainty and nonlinearity of flood, existing hydrological solutions always achieve low prediction robustness while machine learning (ML) approaches neglect the physical interpretability of models. In this paper, we focus on the need for flood forecasting and propose an interpretable Spatio-Temporal Attention Long Short Term Memory model (STA-LSTM) based on LSTM and attention mechanism. We use dynamic attention mechanism and LSTM to build model, Max-Min method to normalize data, variable control method to select hyperparameters, and Adam algorithm to train the model. Emphasis is placed on the visualization and interpretation of attention weights. Experiment results on three small and medium basins in China suggest that the proposed STA-LSTM model outperforms Historical Average (HA), Fully Connected Network (FCN), Convolutional Neural Networks (CNN), Graph Convolutional Networks (GCN), original LSTM (LSTM), spatial attention LSTM (SA-LSTM), and temporal attention LSTM (TA-LSTM) in most cases. Visualization and interpretation of spatial and temporal attention weights reflect the reasonability of the proposed attention-based model. [ABSTRACT FROM AUTHOR]

With rapid development of financial services and products, credit risk assessment has recently gained considerable attention in the field of financial risk management. In this paper, an improved credit risk assessment approach is presented. Based on the credit data from China Banking Regulatory Commission (CBRC), a multi-dimensional and multi-level credit risk indicator system is constructed. In particular, we present an improved sequential minimal optimization (SMO) learning algorithm, named four-variable SMO (FV-SMO), for credit risk classification model. At each iteration, it jointly selects four variables into the working set and an theorem is proposed to guarantee the analytical solution of sub-problem. The assessment is made on China credit dataset and two benchmark credit datasets from UCI database and CD-ROM database. Experimental results demonstrate FV-SMO is competitive in saving the computational cost and outperforms other five state-of-the-art classification methods in credit risk assessment accuracy. [ABSTRACT FROM AUTHOR]