cv

General Information

Full Name Bohui Zhang
Languages English (fluent), Chinese (native)

Skills

Languages Python (very good), Java, JavaScript, MATLAB
Frameworks PyTorch, transformers, KGTK
Semantic Web Tech Stacks OWL, RDF, LOD, SPARQL, PROV
Databases MongoDB, MySQL, Neo4j

Education

  • 2022 - NOW
    PhD, Computer Science
    King's College London, London, UK
    • Advisors: Prof. Elena Simperl, Dr. Albert Meroño Peñuela
  • 2020 - 2021
    M.Sc., Applied Data Science
    University of Southern California, Los Angeles, USA
    • GPA: 3.9/4.0
  • 2017 - 2018
    B.Sc., Materials and Nanosciences
    University of Waterloo, Waterloo, Canada
    • Graduation Dean's Honours List
  • 2016 - 2019
    B.Eng., Nanomaterials and Nanotechnology
    Beijing Jiaotong University, Beijing, China

Experience

  • 2021
    Knowledge Graphs Enrichment Model
    Student Researcher @ Information Sciences Institute, University of Southern California
    • Supervisor: Dr. Filip Ilievski
    • Investigated the feasibility of enriching Wikidata with structured data sources from the linked open data (LOD) cloud.
    • Proposed a method that consists of several steps: gap detection, external graph selection, schema alignment, knowledge retrieval, and validation, implemented the procedure using the Knowledge Graph Toolkit (KGTK).
    • Evaluated the method on enriching Wikidata with two LOD sources: DBpedia and Getty Vocabularies. The experiments showed that the LOD-based method can enrich Wikidata with millions of new high-quality statements efficiently.
  • 2021
    Neural Architecture Search in Real-World Computer Vision Application
    Machine Learning Intern @ Alibaba Cloud, Hangzhou, China
    • Supervisor: Jingjun (Alvin) Chu
    • Worked on a Neural Architecture Search (NAS) system for optimizing models in search space defined by ProxylessNAS, used on image classification and feature extraction tasks based on dataset collecting from group's retail sector.
    • Improved the model training process using knowledge distillation and improved the optimal model architecture searching process in various hardware environments using policy gradient algorithm based on target accuracy, FLOPs and latency.
    • The optimal models deployed on terminal machines achieved model compression for more than 60% decrease on FLOPs while improving rank1 and rank6 compared with state-of-the-art MobileNetV2 models and keeping the top1 accuracy above 98%.

Honors and Awards

  • 2022 - 2024
    King's College London
    • NMES Enterprise & Engagement Partnerships Fund
    • ISWC 2023 Student Travel Award
    • Humane-AI NET Travel Awards
  • 2019
    University of Waterloo
    • Graduation Dean's Honours List
  • 2015 - 2019
    Beijing Jiaotong University
    • Waterloo-Beijing Jiaotong University Tuition Award (2016, 2017, 2018)
    • Excellence Scholarship of Academic Activities (2017 - 2018)
    • Excellence Scholarship of Social Activities (2015 - 2016)

Academic Services

Teaching Assistant

  • 7CUSMNDA Network Data Analysis
  • 5CCS2FC2 Foundations of Computing 2