Browsing by Author "Sun, Yanshen"

Now showing 1 - 5 of 5
  • Thumbnail Image
    Detecting anomalous traffic behaviors with seasonal deep Kalman filter graph convolutional neural networks
    Sun, Yanshen; Lu, Yen-Cheng; Fu, Kaiqun; Chen, Fanglan; Lu, Chang-Tien (Elsevier, 2022-09)
    Anomaly detection over traffic data is crucial for transportation management and abnormal behavior identification. An anomaly in real-world scenarios usually causes abnormal observations for multiple detectors in an extended period. However, existing anomaly detection methods overly leverage the single or isolated feature interdependent contextual information in anomalies, inevitably dropping the detec-tion performance. In this paper, we propose S-DKFN (Seasonal Deep Kalman Filter Network), to identify abnormal patterns with a long duration and wide coverage. S-DKFN models traffic data with a graph and simultaneously investigates the spatial and temporal features to hunt abnormal behaviors. Specifically, a dilation temporal convolutional network (TCN) is used to merge the multi-granular seasonal features and a graph convolution network (GCN) to extract spatial features. The outputs of TCN and GCN are then fed to long-short term models (LSTM) and merged by Kalman filters for denoising. An encoder-decoder mod-ule is introduced to predict traffic attributes with seasonal features. The mean squared errors (MSE) of the predictions are considered the anomaly scores. Experimental results on two real-world datasets show that our proposed S-DKFN framework outperforms the state-of-the-art baseline methods in detecting anomalies with long-duration and wide-coverage, especially its ability to detect accidents.(c) 2022 The Authors. Published by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
  • Thumbnail Image
    Evaluating the quality of ground surfaces generated from Terrestrial Laser Scanning (TLS) data
    Sun, Yanshen (Virginia Tech, 2019-06-24)
    Researchers and GIS analysts have used Aerial Laser Scanning (ALS) data to generate Digital Terrain Models (DTM) since the 1990s, and various algorithms developed for ground point extraction have been proposed based on the characteristics of ALS data. However, Terrestrial Laser Scanning (TLS) data, which might be a better indicator of ground morphological features under dense tree canopies and more accessible for small areas, have been long ignored. In this research, the aim was to evaluate if TLS data were as qualified as ALS to serve as a source of a DTM. To achieve this goal, there were three steps: acquiring and aligning ALS and TLS of the same region, applying ground filters on both of the data sets, and comparing the results. Our research area was a 100m by 140m region of grass, weeds and small trees along Strouble's Creek on the Virginia Tech campus. Four popular ground filter tools (ArcGIS, LASTools, PDAL, MCC) were applied to both ALS and TLS data. The output ground point clouds were then compared with a DTM generated from ALS data of the same region. Among the four ground filter tools employed in this research, the distances from TLS ground points to the ALS ground surface were no more than 0.06m with standard deviations less than 0.3m. The results indicated that the differences between the ground extracted from TLS and that extracted from ALS were subtle. The conclusion is that Digital Terrain Models (DTM) generated from TLS data are valid.
  • Thumbnail Image
    Front-End Kibana (FEK) CS5604 Fall 2019
    Powell, Edward; Liu, Han; Huang, Rong; Sun, Yanshen; Xu, Chao (Virginia Tech, 2020-01-13)
    During the last two decades, web search engines have been driven to new quality levels due to the continuous efforts made to optimize the effectiveness of information retrieval. More and more people are becoming satisfied during their information retrieval processes, and web search has gradually replaced older methods, where people obtained information from each other or from libraries. Information retrieval systems are in constant interaction with users and help users interpret and analyze data. Currently, we are building the front end of a search engine, where users can explore information related to Tobacco Settlement documents from the University of California, San Francisco, as well as the Electronic Theses and Dissertations (ETDs) of Virginia Tech (and possibly other sites). This submission introduces the current work of the front-end team to build a functional user interface, which is one of the key components of a larger project to build a state-of-the-art search engine for two large datasets. We also seek to understand how users search for data, and accordingly provide the users with more insight and utilities from the two datasets with the help of the visualization tool Kibana. Already, a search website, where users can explore the two datasets, Tobacco Settlement dataset and ETDs dataset, has been created. A series of functionalities of the searching page have been realized, for instance, the login system, searching, filter functions, a Q&A page, and a visualization page.
  • Thumbnail Image
    The Scalability of X3D4 PointProperties: Benchmarks on WWW Performance
    Sun, Yanshen (Virginia Tech, 2020-09-29)
    With the development of remote sensing devices, it becomes more and more convenient for individual researchers to acquire high-resolution point cloud data by themselves. There have been plenty of online tools for researchers to exhibit their work. However, the drawback of existing tools is that they are not flexible enough for the users to create 3D scenes of a mixture of point-based and triangle-based models. X3DOM is a WebGL-based library built on Extensible 3D (X3D) standard, which enables users to create 3D scenes with only a little computer graphics knowledge. Before X3D 4.0 Specification, little attention has been paid to point cloud rendering in X3DOM. PointProperties, an appearance node newly added in X3D 4.0, provides point size attenuation and texture-color mixing effects to point geometries. In this work, we propose an X3DOM implementation of PointProperties. This implementation fulfills not only the features specified in X3D 4.0 documentation, but other shading effects comparable to the effects of triangle-based geometries in X3DOM, as well as other state-of-the-art point cloud visualization tools. We also evaluate the performances of some of these effects. The result shows that a general laptop is able to handle most of the examined conditions in real-time.
  • Thumbnail Image
    Spatial Temporal Graph Neural Networks for Decentralized Control of Robot Swarms
    Chen, Siji; Sun, Yanshen; Li, Peihan; Zhou, Lifeng; Lu, Chang-Tien (ACM, 2023-11-13)
    Recent research has explored the use of graph neural networks (GNNs) for decentralized control in swarm robotics. However, it has been observed that relying solely on local states is insufficient to imitate a centralized control policy. To address this limitation, previous studies proposed incorporating đŸ-hop delayed states into the computation. While this approach shows promise, it can lead to a lack of consensus among distant flock members and the formation of small localized groups, ultimately resulting in task failure. Our approach is to include the delayed states to build a spatiotemporal GNN model (ST-GNN) by two levels of expansion: spatial expansion and temporal expansion. The spatial expansion utilizes đŸ-hop delayed states to broaden the network while temporal expansion, can effectively predict the trend of swarm behavior, making it more robust against local noise. To validate the effectiveness of our approach, we conducted simulations in two distinct scenarios: free flocking and flocking with a leader. In both scenarios, the simulation results demonstrated that our decentralized ST-GNN approach successfully overcomes the limitations of local controllers. We performed a comprehensive analysis on the effectiveness of spatial expansions and temporal expansions independently. The results clearly demonstrate that both significantly improve overall performance. Furthermore, when combined, they achieve the best performance compared to global solution and delayed states solutions. The performance of ST-GNN underscores its potential as an effective and reliable approach for achieving cohesive flocking behavior while ensuring safety and maintaining desired swarm characteristics.