Borui Zhang

About me: I am a third year Ph.D student in i-VisionGroup in the Department of Automation, Tsinghua University, advised by Prof. Jiwen Lu. In 2021, I received my BS degree from the Department of Automation and a second BS degree from the School of Economics and Management, THU.

Research: I have a broad interest in computer vision and deep learning theory. At present, my research mainly focus on:

  • Explainable AI: Neural attribution and visualization, alignment between representation and concept, and modular network design
  • Optimization theory: Low-resource optimization, optimization for large models, and convergence analysis
  • Computer vision application: Autonomous driving, 3D reconstruction, and implicit neural representation
  • Interest: I cherish traveling and capturing beautiful moments through photography 📷. Classical music 🎹 has always fascinated me, and I've been devoted to practicing Chopin's compositions lately. Additionally, I excel in sports like badminton 🏸, table tennis 🏓, and running 🏃, which serve as my outlets to combat the pressures of scientific research.

    Email  /  Google Scholar  /  Github  /  Photography

    profile photo
    News
  • 2024-02: Two papers on autonomous driving is accepted to CVPR 2024.
  • 2024-01: One paper on explainable deep networks is accepted to ICLR 2024.
  • 2023-01: One paper on explainable deep networks is accepted to ICLR 2023.
  • 2022-07: One paper on dynamic metric learning is accepted to ECCV 2022.
  • 2022-03: One paper on explainable metric learning is accepted to CVPR 2022.
  • 2021-07: One paper on deep metric learning is accepted to ICCV 2021.
  • Preprints

    * indicates equal contribution

    bort-arxiv Exploring Unified Perspective For Fast Shapley Value Estimation
    Borui Zhang*, Baotong Tian*, Wenzhao Zheng, Jie Zhou, Jiwen Lu
    arXiv, 2023
    [Paper] / [Code]

    This paper analyzes the consistency of existing Shapley value estimators and proposes the simple amortized estimator, SimSHAP. Extensive experiments conducted on tabular and image datasets validate the effectiveness of our SimSHAP, which significantly accelerates the computation of accurate Shapley values.

    Publications

    * indicates equal contribution

    bort-arxiv Path Choice Matters for Clear Attribution in Path Methods
    Borui Zhang, Wenzhao Zheng, Jie Zhou, Jiwen Lu
    International Conference on Learning Representations (ICLR), 2024
    [Paper] / [Code]

    To address the ambiguity in attributions caused by different path choices, we introduced the Concentration Principle and developed SAMP, an efficient model-agnostic interpreter. By incorporating the infinitesimal constraint (IC) and momentum strategy (MS), SAMP provides superior interpretations.

    bort-arxiv Bort: Towards Explainable Neural Networks with Bounded Orthogonal Constraint
    Borui Zhang, Wenzhao Zheng, Jie Zhou, Jiwen Lu
    International Conference on Learning Representations (ICLR), 2023
    [Paper] / [Code]

    This paper proposes Bort, an optimizer for improving model explainability with boundedness and orthogonality constraints on model parameters, derived from the sufficient conditions of model comprehensibility and invertibility.

    clcd-eccv2022 Dynamic Metric Learning with Cross-Level Concept Distillation
    Wenzhao Zheng, Yuanhui Huang, Borui Zhang, Jie Zhou, Jiwen Lu
    European Conference on Computer Vision (ECCV), 2022
    [Paper] / [Code]

    This paper propose a hierarchical concept refiner to construct multiple levels of concept embeddings of an image and them pull closer the distance of the corresponding concepts to facilitate the cross-level semantic structure of the image representations.

    avsl-cvpr2022 Attributable Visual Similarity Learning
    Borui Zhang, Wenzhao Zheng, Jie Zhou, Jiwen Lu
    Conference on Computer Vision and Pattern Recognition (CVPR), 2022
    [Paper] / [Code]

    This paper proposes an attributable visual similarity learning (AVSL) framework, which employs a generalized similarity learning paradigm to represent the similarity between two images with a graph for a more accurate and explainable similarity measure between images.

    drml-iccv2021 Deep Relational Metric Learning
    Wenzhao Zheng*, Borui Zhang*, Jiwen Lu, Jie Zhou
    IEEE International Conference on Computer Vision (ICCV), 2021
    [Paper] / [Code]

    This paper proposes to adaptively learn an ensemble of features that characterizes an image from different aspects to model both interclass and intraclass distributions and employs a relational module to capture the correlations among each feature.

    Honors and Awards
  • 2023 National Scholarship (highest scholarship given by the government of China)
  • 2022 Tsinghua Outstanding Student Cadre Award
  • 2022 Tsinghua Excellent Teaching Assistant Award
  • 2021 Tsinghua Future Scholars Scholarship (GPA: 3.81/4.0, Rank: 1/181)
  • 2020 Changtong Scholarship (highest scholarship for seniors in the Dept. of Automation)
  • 2019 Jiang Nanxiang Scholarship (highest scholarship for juniors in Tsinghua)
  • 2018 National Scholarship (highest scholarship given by the government of China)
  • Academic Services
  • Conference Reviewer:
  • CVPR 2024, ICME 2024
  • ICCV 2023, ICME 2023
  • CVPR 2022, ICME 2022, VCIP 2022
  • Journal Reviewer: T-IP,

  • © Borui Zhang | Last updated: Nov. 8, 2023
    Website template from Jon Barron