Avatar

Qinge Xie

Mobile Systems and Networking Group

Fudan University

About Me

I'm a master student in the School of Computer Science in Fudan University. I am advised by Prof. Yang Chen, and work with Mobile Systems and Networking (MSN) Group. I visited Aalto University as a research intern in the Mobile Cloud Computing Group, supervised by Prof. Yu Xiao in 2018 and visited the University of Chicago as a research intern in Sand Lab, supervised by Prof. Ben Y. Zhao and Prof. Heather Zheng in 2019.

My research interests are primarily in systems and security, particularly in mobile systems and computing, security problems for system, deep learning and networking, and applications of deep learning to systems. See my CV here. (Email: qgxie17@fudan.edu.cn)

Education

 
 
 
 
 

Fudan University

M.S. in Computer Science

Sep 2017 – Present Shanghai, China
  • Mobile Systems and Networking Group

  • Huawei Scholarship

 
 
 
 
 

Zhejiang University of Technology

B.S. in Computer Science

Sep 2013 – Jun 2017 Hangzhou, China
  • Comprehensive Rank: 1/27

  • National Scholarship

  • Top Ten Outstanding Undergraduates

Publications

Trimming Mobile Applications for Bandwidth-Challenged Networks in Developing Regions.
IEEE Transactions on Mobile Computing, 2019. (Under Review)
'How do urban incidents affect traffic speed?' A Deep Graph Convolutional Network for Incident-driven Traffic Speed Prediction.
Main technical tracks of ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) 2020. (Under Review)
LBSLab: A User Data Collection System in Mobile Environments.
MHC workshop, ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) 2018. (Accept)
Understanding user activity patterns of the swarm app: A data-driven study.
AppLens workshop, ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) 2018. (Accept)

Research Experience

An Ongoing Work of Adversarial Attack on Graph Data of Real Scenes

  • Proposed a black-box system design of adversarial attack on graph data of real scenes (node classification task).
  • Built an attack model based on Graph Convolutional Network model and implemented a similarity-based attack method. Implemented large-scale training (node sampling) on large-scale graph, e.g., Reddit dataset.

Image Based Renovation Progress Estimation

  • Built a Visual Geometry Group(VGG) based deep learning model for predicting the renovation progress of kitchen images (multiple classification). Achieved a 0.435 test Top1-accuracy and a 0.85 test Top2 accuracy on kitchen dataset.
  • Built a VGG based deep learning model to evaluate the quality of renovation (binary classification). Achieved a test 0.91 accuracy on kitchen dataset.
  • Implemented an occlusion sensitivity method to detect the key areas in images that affect the prediction results.

#1 Incident-driven Real-time Traffic Speed Prediction

  • Collected multi-sources urban traffic data of San Francisco and New York City and performed data processing and analysis.
  • Proposed an urban critical incident discover method and designed a binary classifier to extract the latent impact features of traffic incidents for improving speed prediction. Achieved a 0.82 test F1-score of SFO and a 0.80 test F1-score of NYC.
  • Proposed a Deep Graph Convolutional Network to effectively incorporate incident, spatio-temporal, periodic and context features for traffic speed prediction. Achieved a 0.82 test F1-score of SFO and a 0.80 test F1-score of NYC. Achieved a 11.02% Mean Absolute Percentage Error(MAPE) of SFO and 17.21% MAPE of NYC.

#2 Trimming Mobile Applications for Bandwidth-Challenged Networks

  • Implemented a WeChat mini-program with identical functionality as an existing Android app to understand sources of app size discrepancy. Performed a empirical analysis of 200 mini-programs and their Android counterparts.
  • Crawled and decompiled 3200 Android apps. Performed detailed analysis and confirmed linked libraries as a dominant factor in apps's overall size.
  • Developed an app trimming framework to automatically trim existing Android apps. For 40% of the test apps, the framework can reduce the app size by at least 10MB.

#3 Mining and Modeling User Behavior in Online Social Networks

  • Applied network packet capture to hack the communication protocol of a widely used Location-Based Social Application (LBSA) Swarm and collected more than 33 million check-ins of 20 thousand users. Performed data processing and analysis.
  • Built a machine learning based model for predicting travelers's preferences for check-in venue types.

#4 LBSLab: A User Data Collection System in Mobile Environments

  • leaded the front-end development of the system. Designed and developed several representative location related functions, e.g., conducting check-ins.
  • Introduced the asynchronous programming pattern to the front end to reduce the latency and leveraged client-based cache to reduce the network traffic.

Selected Awards

Huawei Scholarship

National Scholarship

Top Ten Outstanding Undergraduates

The First Prize of Group Programming Ladder Tournament in China Collegiate Computing Contest

The First Prize of Collegiate Programming Contest in Zhejiang Province

Silver Medal of ACM International Collegiate Programming Contest (ACM-ICPC), Asian Regional

Silver Medal of ACM International Collegiate Programming Contest (ACM-ICPC), Asian Regional