Yingjun Du

Yingjun Du

杜英军

PhD Student

University of Amsterdam, AIM Lab

About me

I am currently a final-year Ph.D. candidate at University of Amsterdam (UvA) and also a member of ELLIS society, working on Meta-learning. I am grateful to be supervised by Cees Snoek. I also received guidance from Xiantong Zhen. Prior to that, I completed my master’s degree in Software Engineering at Beihang University, China. My research interests are: meta-learning, few-shot learning, diffusion model, and variational inference.

I'm on the job market. Feel free to contact me. [Meta-learning Expert]

News

  • [Sep 2023] One paper accepted to NeurIPS 2023.
  • [June 2023] One paper accepted to CoLLAs 2023. (top 12 papers.)
  • [Apr 2023] One paper accepted to ICML 2023.
  • [Feb 2023] One paper accepted to CVPR 2023.
  • [July 2022] Attending International Gran Canaria School on Deep Learning.
  • [May 2022] One paper accepted to TPAMI.
  • [Jan 2022] One paper accepted to ICLR 2022.
  • [Dec 2021] One paper accepted to ACL 2021.
  • [Jan 2021] One paper accepted to ICLR 2021.
  • [Sep 2020] One paper accepted to NeurIPS 20.
  • [July 2020] One paper accepted to ECCV 20.
  • [June 2020] One paper accepted to ICML 20.

Publications

2023

ProtoDiff: Learning to Learn Prototypical Networks by Task-Guided Diffusion
NeurIPS 2023
We introduce ProtoDiff, a novel framework that leverages a task-guided diffusion model during the meta-training phase to gradually generate prototypes, thereby providing efficient class representations.
Yingjun Du, Zehao Xiao, Shengcai Liao, Cees Snoek
PDF Cite Code
ProtoDiff: Learning to Learn Prototypical Networks by Task-Guided Diffusion
EMO: Episodic Memory Optimization for Few-Shot Meta-Learning
CoLLAs 2023 (top 12 papers.)
We propose an episodic memory optimization for meta-learning, we call EMO, which is inspired by the human ability to recall past learning experiences from the brain's memory.
Yingjun Du, Jiayi Shen, Xiantong Zhen, Cees Snoek
PDF Cite Code
EMO: Episodic Memory Optimization for Few-Shot Meta-Learning
MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks
ICML 2023
We propose a method for few-shot learning with fewer tasks, which we call MetaModulation.
Wenfang Sun, Yingjun Du, Xiantong Zhen, Fan Wang, Ling Wang, Cees Snoek
PDF Cite Code
MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Taskses
SuperDisco: Super-Class Discovery Improves Visual Recognition for the Long-Tail
CVPR 2023
We propose SuperDisco to discover super-class representations for long-tailed recognition using a graph model.
Yingjun Du, Jiayi Shen, Xiantong Zhen, Cees Snoek
PDF Cite Code
SuperDisco: Super-Class Discovery Improves Visual Recognition for the Long-Tail

2022

Hierarchical Variational Memory for Few-shot Learning Across Domains
ICLR 2022
We propose a hierarchical memory that stores features at different semantic levels for the cross-domain few-shot learning.
Yingjun Du, Xiantong Zhen, Ling Shao, Cees Snoek
PDF Cite Code
Hierarchical Variational Memory for Few-shot Learning Across Domains
MetaKernel: Learning Variational Random Features with Limited Labels
TPAMI
We propose meta-learning kernels with random Fourier features for few-shot learning.
Yingjun Du, Haoliang Sun, Xiantong Zhen, Jun Xu, Yilong Yin, Ling Shao, Cees Snoek
PDF Cite Code
MetaKernel: Learning Variational Random Features with Limited Labels

2021

Meta-Learning with Variational Semantic Memory for Word Sense Disambiguation
ACL 2021
We propose a model of semantic memory for word sense disambiguation in a meta-learning setting.
Yingjun Du, Nithin Holla, Xiantong Zhen, Cees Snoek Ekaterina Shutova
PDF Cite Code
Meta-Learning with Variational Semantic Memory for Word Sense Disambiguation
MetaNorm: Learning to Normalize Few-shot Batches Across Domains
ICLR 2021
We propose MetaNorm, a simple yet effective meta-learning normalization. It tackles the aforementioned issues in a unified way by leveraging the meta-learning setting and learns to infer adaptive statistics for batch normalization.
Yingjun Du, Xiantong Zhen, Ling Shao, Cees Snoek
PDF Cite Code
MetaNorm: Learning to Normalize Few-shot Batches Across Domains
Learning to Adapt With Memory for Probabilistic Few-Shot Learning
TCSVT 2021
We propose a new probabilistic framework that learns to fast adapt with external memory.
Lei Zhang, Liyun Zuo, Yingjun Du, Xiantong Zhen
PDF Cite Code
Learning to Adapt With Memory for Probabilistic Few-Shot Learning
Variational Abnormal Behavior Detection With Motion Consistency
TIP 2021
We propose a new probabilistic framework named variational abnormal behavior detection (VABD), which can detect abnormal crowd behavior in video sequences..
Jing Li, Qingwang Huang, Yingjun Du, Xiantong Zhen, Shengyong Chen, Ling Shao
PDF Cite Code
Variational Abnormal Behavior Detection With Motion Consistency

2020

Learning to Learn Variational Semantic Memory
NeurIPS 2020
We introduce variational semantic memory into meta-learning to acquire long-term knowledge for few-shot learning.
Xiantong Zhen, Yingjun Du, Cees Snoek, Ling Shao
PDF Cite Code
Learning to Learn Variational Semantic Memory
Learning to Learn Kernels with Variational Random Features
ICML 2020
We introduce kernels with random Fourier features in the meta-learning framework for few-shot learning.
Xiantong Zhen, Yingjun Du, Haoliang Sun, Jun Xu, Yilong Yin, Ling Shao, Cees Snoek
PDF Cite Code
Learning to Learn Kernels with Variational Random Features
Learning to Learn with Variational Information Bottleneck for Domain Generalization
ECCV 2020
To deal with domain shift, we learn domain-invariant representations by the proposed principle of meta variational information bottleneck, we call MetaVIB.
Yingjun Du, Xiantong Zhen, Jun Xu, Huan Xiong, Qiang Qiu, Ling Shao, Cees Snoek
PDF Cite Code
Learning to Learn with Variational Information Bottleneck for Domain Generalization
Variational Image Deraining
WACV 2020
We propose a variational image deraining (VID) method by formulating image deraining in a conditional variational auto-encoder framework.
Yingjun Du, Jun Xu, Qiang Qiu, Xiantong Zhen, Lei Zhang
PDF Cite Code
Variational Image Deraining
Conditional Variational Image Deraining
TIP 2020
We propose a Conditional Variational Image Deraining (CVID) network for better deraining performance, leveraging the exclusive generative ability of Conditional Variational Auto-Encoder (CVAE) on providing diverse predictions for the rainy image.
Yingjun Du, Jun Xu, Xiantong Zhen, Ming-Ming Cheng, Ling Shao
PDF Cite Code
Conditional Variational Image Deraining

Academic Activities

  • Conference reviewer: NeurIPS 21-23, ICML 21-22, ICLR 20-23, CVPR 21-22, ICCV 23.
  • Journal reviewer: Pattern Recognition, TPAMI, TCSVT.
  • Teaching Assistant: Computer vision (2021), Deep learning vision (2021), Apply machine learning (2022, 2023),

    • Contact

    • y.du@uva.nl
    • Science Park 900, LAB42, Amsterdam, 1098XH