Close

Ring Xiaoyuan Liang

Ph.D. Candidate in Computer Science

Research in deep learning, data mining and intelligent transportation

Email: XL367@njit.edu

Download CV

News

Experience

NEC Labs America

Data scientist intern

  • Implemented a supervised and unsupervised deep learning model based on attention and LSTM to detect anomaly events on the multiple sensors' data using Tensorflow
  • Designed and implemented a database system to store the trained model and model results using ElasticSearch
  • Visualized the detection results in a website with the help of Grafana

NEC Labs America

Data scientist intern

  • Modeled the pairwise relation between events using pairwise Hawkes processes to detect anomaly events using Sklearn
  • Adapted a pairwise structure using a convolutional neural network on similarity tensors to detect anomaly events
  • Extracted features from bank transaction data and built unsupervised models with DBSCAN to detect anomaly users

University of Houston

Research assistant

  • Proposed a dynamic traffic light control system to minimize vehicles' overall waiting time based on reinforcement learning
  • Modeled the phases' time changes in a cycle of a traffic light as a high-dimension Markov decision process and made the high-dimension decisions as the actions
  • Proposed a double dueling deep Q learning network to match states and Q values considering illegal actions and trained the network using prioritized experience replay
  • Reduced over 20% vehicles’ average time from the starting point via simulation in a multi-lane intersection in SUMO and built our model to control the traffic light using TensorFlow

Education

New Jersey Institute of Technology

Sept 2013 - Current

Ph.D. in Computer Science

Harbin Institute of Technology

Sept 2009 - Sept 2013

Bachelor of Engineering in Information Security at Computer Science and Technology Department

Publications

[ Google scholar ]

Sort by Area(Default) / Date

Deep learning/machine learning

  1. Learning K-way D-dimensional Discrete Embedding for Hierarchical Data Visualization and Retrieval [TBD]
    Xiaoyuan Liang, Martin Renqiang Min, Hongyu Guo, Guiling Wang
    [IJCAI-19]The 28th International Joint Conference on Artificial Intelligence, 2019. (Acceptance Rate: 805/4752 = 17.89%)
  2. A Deep Spatio-Temporal Fuzzy Neural Network for Passenger Demand Prediction [PDF]
    Xiaoyuan Liang, Guiling Wang, Martin Renqiang Min, Qi Yi, Zhu Han
    [SDM19]The SIAM International Conference on Data Mining, 2019. (Acceptance Rate: 90/397 = 22.7%)
  3. A Deep Learning Model for Transportation Mode Detection Based on Smartphone Sensing Data []
    Xiaoyuan Liang, Yuchuan Zhang, Guiling Wang, Songhua Xu
    [IEEE TITS] IEEE Transactions on Intelligent Transportation Systems, under submission.
  4. A Deep Reinforcement Learning Network for Traffic Light Cycle Control [PDF]
    Xiaoyuan Liang, Xunsheng Du, Guiling Wang, Zhu Han
    [IEEE TVT] IEEE Transactions on Vehicular Technology, Vol. 68, No. 2, pp. 1243-1253, 2019. (IF2018=4.432)
  5. A Low-Cost Collaborative Indoor Localization System based on Smartphone Platform [PDF]
    Xiaoyuan Liang, Guiling Wang, Zhu Han
    The 14th IEEE International Conference on Green Computing and Communications (GreenCom), 2018.
  6. A Convolutional Neural Network for Transportation Mode Detection Based on Smartphone Platform [PDF]
    Xiaoyuan Liang, Guiling Wang
    [MASS]The 14th IEEE International Conference on Mobile Ad-hoc and Sensor Systems, pp. 338-342, 2017.
  7. A Risk and Similarity Aware Application Recommender System [PDF]
    Xiaoyuan Liang, Jie Tian, Xiaoning Ding, Guiling Wang
    [CIT] Journal of Computing and Information Technology, Vol. 23(4) pp. 303-315, 2015.

IoT/Wireless sensor networks

  1. A Distributed Intersection Management Protocol for Safety, Efficiency, and Driver's Comfort [PDF]
    Xiaoyuan Liang, Tan Yan, Joyoung Lee, Guiling Wang
    [IEEE IoT] IEEE Internet of Things Journal, Vol. 5, No. 3, pp. 1924-1935, 2018. (IF2018=5.863)
  2. WA-MAC: A Weather Adaptive MAC Protocol in Survivability-Heterogeneous Wireless Sensor Networks [PDF]
    Jie Tian, Yi Wang, Xiaoyuan Liang, Guiling Wang, Yujun Zhang
    [ADHOC] Elsevier Ad Hoc Networks, Vol. 67, pp. 40-52, 2017. (IF2017=3.151)
  3. Deployment and Reallocation in Mobile Survivability-Heterogeneous Wireless Sensor Networks for Barrier Coverage [PDF]
    Jie Tian, Xiaoyuan Liang, Guiling Wang
    [ADHOC] Elsevier Ad Hoc Networks, Vol. 36 pp. 321-331, 2016. (IF2016=3.047)
  4. A Novel Set Division Algorithm for Joint Scheduling and Routing in Wireless Sensor Networks [PDF]
    Jie Tian, Xiaoyuan Liang, Tan Yan, Mahesh Kumar Somashekar, Guiling Wang, Cesar Bandera
    [WINET] Springer Wireless Networks, Vol. 21, pp. 1443-1455, 2015.
  5. ARPP: An Augmented Reality 3D Ping-Pong Game System on Android Mobile Platform
    Xin Gao, Jie Tian, Xiaoyuan Liang, Guiling Wang
    The 23rd IEEE Wireless and Optical Communication Conference (WOCC), pp. 1-6, 2014.

Poster

  1. A Light-weight and Accurate Transit Mode Detection System Based on Smartphone Platform [PDF]
    Xiaoyuan Liang, Guiling Wang
    Transportation Research Board 96th Annual MeetingTransportation Research Board, 2017.
  2. Deploying Mobile Survivability-Heterogeneous Sensor Networks for Barrier Coverage [PDF]
    Xiaoyuan Liang, Jie Tian, Guiling Wang
    2014 IEEE North Jersey Advanced Communications Symposium (NJACS), 2014.

Projects

Time-series Anomaly Events Visualization

June 2018 – August 2018

  • Detected the anomaly events of the current period's data based on deep learning and calculated the anomaly scores
  • Implemented a website to display the anomaly events shifted by time changes using Grafana
  • Displayed the nearest neighbors, neighbors' data and alerts of the current time period in one website

Deep Spatio-Temporal Fuzzy Network for Passenger Demand Prediction

January 2018 – June 2018

  • Designed an end-to-end neural network with fusing heterogeneous data to predict passenger demands for a transportation network company
  • Combined deep learning, including ConvLSTM and attention, and fuzzy neural network to capture different feature interactions
  • Achieved more than 10% improvement over the state-of-the-art methods in experiments on Keras

Deep Reinforcement Learning for Traffic Light Control

June 2017 – September 2017

  • Proposed a dynamic traffic light control system to minimize vehicles' overall waiting time based on deep reinforcement learning
  • Proposed a double dueling deep Q learning network to match states and Q values considering illegal actions and trained the network using prioritized experience replay
  • Reduced over 20% vehicles’ average time from the starting point of training via simulation in a multi-lane intersection in SUMO and TensorFlow

Deep Learning for Transportation Mode Detection

December 2016 – May 2017

  • Detected users' seven transportation modes using only the smartphones' accelerometer data
  • Preprocessed the data to reduce the influence of meaningless fluctuation and the influence of smartphones' rotation and placement
  • Adopted a convolutional neural network on one-dimension data to detect users' transportation modes

A Collaborative Indoor Localization System

January 2016 - March 2017

  • Proposed a framework to predict smartphones' locations using the collected ambient sound from each pair of smartphones
  • Preserved the energy consumption via waking smartphones when detecting the sound activities using a Gaussian model on the signal-to-noise ratio
  • Recognized sound activities of different sources from the time-series sound using HMM and GMM
  • Calculated time difference of two sounds by cross-correlation and smartphones’ position based the time difference using triangulation

Distributed Vehicle Management System at Intersections base on VANETs

September 2015 - December 2016

  • Proposed and designed a vehicle management system at intersections in a distributed way via VANETs
  • Guaranteed the drivers' safety and the traffic efficiency at intersections in many metrics
  • Simulated the distributed traffic scenario at an isolated intersection using C++ in OMNET++ and SUMO

Barrier Coverage in Heterogeneous WSNs

September 2014 - January 2016

  • Proposed and solved a new barrier connection problem in harsh environment when only some sensors are robust and movable
  • Proposed a greedy barrier construction algorithm to guarantee the maintainance in both normal and harsh environment

Traffic Information Website

June 2015 - August 2015

  • Implemented a website to display traffic information of routes generated by Google Maps APIs automatically
  • Integrated a website with two modes for the operator, which is protected by passwords, and viewers, separately
  • Achieved the website's dynamic view with Ajax, JavaScript and MySQL.

Risk and Similarity Aware Mobile APP Recommender

September 2014 - September 2015

  • Proposed two algorithms to estimate an application's potential risk based on the requested permissions and to calculate the similarity between two applications via the permissions, categories, and reviews
  • Designed a personalized Android application recommender system based on the awareness of the similarity to the installed applications and the applications' potential risks

Professional Activities

Reviewer

  • IEEE Transactions on Vehicular Technology
  • IEEE Transactions on Mobile Computing
  • IEEE Signal Processing Magazine
  • IEEE Access
  • IEEE Transactions on Computational Social Systems
  • 2019 ACM SIGKDD Conference on Knowledge Discovery and Data Mining
  • 2019 International Conference on Computing, Networking and Communications (ICNC): Machine Learning for Communication and Networking
  • NIPS 2018 Machine Learning for Intelligent Transportation Systems Workshop
  • 2016 IEEE International Performance Computing and Communications Conference

Teaching assistant

  • CS677 Deep learning
  • CS675 Introduction to Machine learning
  • CS656 Internetworking & Higher Layer Protocol
  • CS634 Cloud Computing
  • CS113 Introduction to Computer Science I (C++)
  • CS113 Introduction to Computer Science I (Java)

Skills

Get in Touch