Zheye Deng

E-mail: zdengah at cse dot ust dot hk

Hi! I am Zheye Deng (邓哲也). I am a second-year Ph.D. student at The Hong Kong University of Science and Technology advised by Prof. Yangqiu Song. I received my B.Sc. in Computer Science from Peking University, where I was advised by Prof. Ming Zhang.

My research interest mainly focus on the integration of graph structures and natural language in machine learning, including the construction and denoising of large-scale commonsense knowledge graphs, as well as the development of neural graph database.

Interests
  • Commonsense Knowledge Graph
  • Logical Rule Mining
  • Neural Graph Database
Education

  • Ph.D. in Computer Science and Engineering, 2022-Present

    The Hong Kong University of Science and Technology

  • B.Sc. in Computer Science and Technology, 2016-2020

    Peking University

Experience

 
 
 
 
 
Research and Development Engineer
Megvii Technology Co., Ltd.
July 2020 – August 2022 Beijing, China
  • Contributed code to MegEngine, a deep learning framework with auto-differentiation.
  • Focused on optimizing the memory usage during model training with SOTA approaches.
  • Improved and implemented a gradient checkpointing algorithm DTR in MegEngine.
 
 
 
 
 
Research Assistant
Dlib Group, Peking University
January 2017 – July 2020 Beijing, China
  • Advisor: Prof. Ming Zhang
  • Focused on natural language processing, recommendation system and data mining:
    • Graph Neural Networks for Table-based Fact Verification
    • Sequential Recommendation with Contrastive Predictive Coding
    • ACM International Conference on Web Search and Data Mining Cup 2018
 
 
 
 
 
Research Assistant
University of California, Santa Barbara
June 2019 – September 2019 Santa Barbara, CA, USA
  • Advisor: Prof. William Wang
  • Published a large and high-quality dataset called AnchorNER using abstracts from Wikipedia and DBpedia
  • Developed a neural correction model for open-domain Named Entity Recognition and obtained SOTA results
 
 
 
 
 
Research and Development Engineer
Bytedance Technology Co., Ltd.
June 2018 – October 2018 Beijing, China
  • Developed an algorithm based on a decision tree to identify comment baiting videos in TikTok.
  • Collaborated with team members using Kafka and HiveQL to process massive user data.

Featured Publications

Zheye Deng, Weiqi Wang, Zhaowei Wang, Xin Liu, Yangqiu Song
October, 2023 In Findings of EMNLP 2023
Gold: A Global and Local-aware Denoising Framework for Commonsense Knowledge Graph Noise Detection
Gold: A Global and Local-aware Denoising Framework for Commonsense Knowledge Graph Noise Detection

We propose Gold, a denoising framework for CSKGs that incorporates entity semantic information, global rules, and local structural information from the CSKG. Experiment results demonstrate that Gold outperforms all baseline methods in noise detection tasks on synthetic noisy CSKG benchmarks. Furthermore, we show that denoising a real-world CSKG is effective and even benefits the downstream zero-shot commonsense question-answering task.
PDF Cite Code

Publications

Zheye Deng, Weiqi Wang, Zhaowei Wang, Xin Liu, Yangqiu Song (2023). Gold: A Global and Local-aware Denoising Framework for Commonsense Knowledge Graph Noise Detection. In Findings of EMNLP 2023.

PDF Cite Code

Zhongzhe Hu, Junmin Xiao, Zheye Deng, Mingyi Li, Kewei Zhang, Xiaoyang Zhang, Ke Meng, Ninghui Sun, Guangming Tan (2022). MegTaiChi: Dynamic tensor-based memory management optimization for DNN training. In ICS 2022.

PDF Cite

Mengdi Zhu, Zheye Deng, Wenhan Xiong, Mo Yu, Ming Zhang, William Yang Wang (2019). Towards open-domain named entity recognition via neural correction models.

PDF Cite Code

Awards

  • Hong Kong PhD Fellowship (2022 - 2026)
  • HKUST RedBird PhD Scholarship (2022)
  • Third-class Scholarship from Peking University (2019)

Teaching

  • COMP4332 Big Data Mining and Management Teaching Assistant (Spring 2023)
  • Learning Python with Data Mining: Experience and Outlook (Spring 2019)
  • Foundations of Computer (Fall 2018)