28 June 2019 in Journal of Transportation Safety & Security
Research efforts have been made to understand tunnels’ traffic safety. Most of the previous studies have not considered the different features of tunnels with different ranges of length comprehensively. Generally, three- or four-zone approach has been adopted, with which the entrance and exit parts of a tunnel are considered symmetrical in the safety analysis. This study employs a seven-zone analytic approach for the safety investigation of 18 expressway tunnels with length ranging from 2 to 3 km. The results reveal that the crash rate increases firstly for the entrance zone, then decreases for the midzone and again increases at the exit zone. The high crash rates at the access, entrance, and transition zones are attributed to rear-end crashes. Although single-vehicle crashes take place in the mid and exit zones (1) at the tunnel entrance: failure to maintain safe distance; (2) in the midarea: failure to maintain safe distance, fatigue driving, overspeeding, and improper lane change; and (3) at the tunnel exit: overspeeding and improper lane change mainly contribute to the crash occurrence. Friedman test was performed to test the significance of the contributing factors. The crash occurrence mechanism is discussed for the selected long tunnels. Finally, engineering and policy countermeasures are recommended to improve traffic safety in expressway tunnels.
6 February 2023 in PLOS ONE
Having observed that gene expressions have a correlation, the Library of Integrated Network-based Cell-Signature program selects 1000 landmark genes to predict the remaining gene expression value. Further works have improved the prediction result by using deep learning models. However, these models ignore the latent structure of genes, limiting the accuracy of the experimental results. We therefore propose a novel neural network named Neighbour Connection Neural Network(NCNN) to utilize the gene interaction graph information. Comparing to the popular GCN model, our model incorperates the graph information in a better manner. We validate our model under two different settings and show that our model promotes prediction accuracy comparing to the other models.
1 August 2017 in Journal of Physics: Conference Series
In order to solve face alignment more effectively, we proposed a multi-task deep convolution network for face alignment, which achieves good performance even in the case of large pose variations and severe occlusion. Instead of dealing with face alignment as a single task, we jointly trained the auxiliary task of pose estimation together with face alignment to guide the distribution of facial points. By doing so we are able to 1) avoid trapping in the local optimum due to the inaccurate face boxes, 2) improve the robustness in dealing with faces with pose variation and severe occlusion. Compared with the traditional methods, our method also improves the accuracy by providing better initialization instead of mean shape. Extensive experiments show that our method has great performance on various benchmarks.