John Thorpe

PhD Student
Computer Science Department
University of California, Los Angeles
Email: jothor@cs.ucla.edu

About Me

I am a PhD student studying Computer Science at UCLA with my advisor Dr. Harry Xu. I got a BS in Computer Science and a BA in Economics from UC Irvine in 2017. My current CV can be found here.

Research

    Current Work:

      Currently I am developing a system to enable scalable and efficient deep neural network learning on graphs, known as Graph Neural Networks. I plan to leverage the extensive research done on scalable graph systems, as well as my own expertise working with graph systems to overcome the current scalability limitations inherent in graph learning. With this project, I also hope to explore the emerging architecture of serverless computing to enable effortless and affordable scalability for computation intensive tasks.

    Grapple: A Graph System for Static Finite-State Property Checking of Large-Scale Systems Code

      Grapple is a system for scalable static analysis of large-scale systems code. In this work, we utilized symbolic execution as a way of enabling path-sensitive error-checking, allowing us to increase the precision of the analysis compared to other techniques. We made the system using out-of-core graph partitioning in order to enable scalability from a single node.

Publications

  • Zhiqiang Zuo, John Thorpe, Yifei Wang, Qiuhong Pan, Shenming Lu, Kai Wang, Guoqing Harry Xu, Linzhang Wang, and Xuandong Li. Grapple: A Graph System for Static Finite-State Property Checking of Large-Scale Systems Code. In Proceedings of the Fourteenth EuroSys Conference 2019 (EuroSys'19). [ PDF ]
  • Kai Wang, Zhiqiang Zuo, John Thorpe, Tien Quang Nguyen, and Guoqing Harry Xu. RStream: Marrying Relational Algebra with Streaming for Efficient Graph Mining on a Single Machine. In Proceedings of the 12th USENIX conference on Operating Systems Design and Implementation (OSDI'18). [ PDF ]