About Me

I am Deyu Tian, an AI platform engineer at China Mobile Research Institute. I received my Ph.D. in Software Engineering from Peking University in 2024, under the guidance of Professors Yun Ma, Xuanzhe Liu, and Gang Huang at the OS Lab of the Software Institute. Prior to my doctoral studies, I earned a Bachelor of Science degree in Computer Science from Peking University in 2018. My research interests encompass distributed systems and deep learning infrastructures.

Education

Ph.D. in Software Engineering

OS Lab, Software Institute, School of Computer Science, Peking University
2018 ~ 2024

I earned my Ph.D. in Software Engineering from Peking University's OS Lab, under the guidance of Professors Yun Ma and Gang Huang. My research focused on Web ecosystems and deep learning infrastructures.

From 2018 to 2022, I served as a Teaching Assistant for Introduction to Computing (A). Additionally, I was a Teaching Assistant for Scientific Paper Writing in 2021.

I was honored with the Award for Scientific Research by Peking University in 2019 and the Award for Academic Excellence by Peking University in 2021.

Bachelor of Science in Computer Science

School of Electronic Engineering and Computer Science, Peking University
2014 ~ 2018

I enrolled in the School of Electronic Engineering and Computer Science (EECS) at Peking University in 2014, earning a Bachelor of Science degree in Computer Science and Technology in 2018.

In 2016, I served as Vice President of the Students’ Computer Association at the School of EECS, where I contributed to the organization of the 2016 ACM-ICPC Asia Beijing Regional Contest.

I was honored to receive the Merit Student Award and Founder Scholarship from Peking University in 2016.

Awards

2021

Award for Academic Excellence

Peking University

2019

Award for Scientific Research

Peking University

2016

Founder Scholarship

Peking University

Merit Student

Peking University

2013

Bronze Medal of National Olympics in Informatics

China Computer Foundation

Projects

Only some selected projects are introduced here.

DNN inference warm-up on GPU engines for Web apps

I propose WPIA, a DNN inference framework that precompiles GPU programs on the server-side to significantly reduce DNN inference warm-up time in Web browsers. Evaluation results demonstrate that WPIA can accelerate DNN warm-up by an average of 4.27x and up to 21.27x, representing a substantial performance improvement.

This work has been published in FCS 2024.

A heterogeneous DNN inference engine for Web apps

I propose WEngine, a heterogeneous DNN engine designed to optimize DNN inference throughput in Web browsers through pipeline parallelism. Evaluation results demonstrate that WEngine can improve DNN inference throughput by an average of 1.91x and up to 3.40x.

This work has been published in MobiSys 2022.

An empirical study of embedded Web pages in mobile apps

We conducted an empirical study to investigate the prevalence, characteristics, and performance of embedded Web pages within mobile applications. To facilitate this analysis, we developed an automated testing tool specifically designed to target these embedded Web pages.

This work has been published in IEEE TMC 2021.

Publications

2024

[FCS '24] Deyu Tian, Yun Ma, Yudong Han, Qi Yang, Haochen Yang, Gang Huang. WPIA: Accelerating DNN Warm-up in Web Browsers by Precompiling WebGL Programs. [DOI] [BibTeX]
[WWW '24] Weichen Bi, Yun Ma, Yudong Han, Yifan Chen, Deyu Tian, Jiaqi Du. FusionRender: Harnessing WebGPU's Power for Enhanced Graphics Performance on Web Browsers. [DOI] [BibTeX]

2023

[ICWS '23] Haiyang Shen, Yun Ma, Yue Li, Xiaoling Wang, Deyu Tian, Tong Jia, Tengfei He, Shenghua Luo. ADPal: Automatic Detection of Troubled Users in Online Service Systems via Page Access Logs. [DOI] [BibTeX]
[WWW '23] Weichen Bi, Yun Ma, Deyu Tian, Qi Yang, Mingtao Zhang, Xiang Jing. Demystifying Mobile Extended Reality in Web Browsers: How Far Can We Go? [DOI] [BibTeX]

2022

[IEEE TMC '22] Deyu Tian, Yun Ma, Aruna Balasubramanian, Yunxin Liu, Gang Huang, Xuanzhe Liu. Characterizing embedded Web browsing in mobile apps. [DOI] [BibTeX]
[MobiSys '22] Deyu Tian, Haiyang Shen, Yun Ma. Parallelizing DNN inference in mobile Web browsers on heterogeneous hardware. [DOI] [BibTeX]

2021

[ICSE-C '21] Deyu Tian. Detecting user-perceived failure in mobile applications via mining user traces. [DOI] [BibTeX]

2019

[Internetware '19] Deyu Tian, Yun Ma. Understanding quality of experiences on different mobile browsers. [DOI] [BibTeX]
[WWW '19] Yun Ma, Dongwei Xiang, Shuyu Zheng, Deyu Tian, Xuanzhe Liu. Moving deep learning into web browser: How far can we go? [DOI] [BibTeX]

2018

[BigComp '18] Junzhi Gong, Deyu Tian, Dongsheng Yang, Tong Yang, Tuo Dai, Bin Cui, Xiaoming Li. SSS: An accurate and fast algorithm for finding top-k hot items in data streams. [DOI] [BibTeX]

2017

[ICC '17] Dongsheng Yang, Deyu Tian, Junzhi Gong, Siang Gao, Tong Yang, Xiaoming Li. Difference bloom filter: A probabilistic structure for multi-set membership query. [DOI] [BibTeX]