Yikai Wang

Ph.D. candidate of Statistics
School of Data Science
Fudan University
Shanghai, China

Links
ORCID
GitHub
Scholar
Outlook Email: yi-kai.wang

About


I am a Ph.D. candidate under the supervision of Prof. Yanwei Fu in the School of Data Science at Fudan University. Prior to this, I completed my Bachelor of Science in Mathematics from Fudan University.

My research primarily revolves around the development of machine learning algorithms that possess both theoretical guarantees and empirical efficacy. Specifically, I am interested in statistical sparsity for learning problems, with a focus on sample selection and applications in few-shot learning (ICI, CVPR20, TPAMI21) as well as learning with noisy labels (SPR, CVPR22). Furthermore, I explore the false selection issue in sample selection and propose a label-knockoff method (Knockoffs-SPR, preprint) to control the false selection rate.

Additionally, I engage in research of unlocking the potential of foundation models, with a focus on using generative decoding models for standard vision tasks like text-guided (SeMani, preprint) and reference-guided (PGIC, preprint) image manipulation, as well as scientific problems such as neural coding (LEA, preprint) and weather forecasting (top 3%, 19/697, on WiDS Datathon 2023).

Apart from my own research, I assist junior students in intriguing research projects. These can be found in miscellaneous publications, including oracle character recognition (FFD Augmentor, an undergraduate project) and so on.

Talks


Sparse Learning for Noisy Data Detection.
[Youtube][Bilibili]
In CVPR 2022 tutorial: Sparse Learning in Neural Networks and Robust Statistical Analysis.

Publications


Note: Asterisk (*) indicates equal contribution and dagger (†) indicates corresponding author(s).

Sparsity for Learning

Knockoffs-SPR: Clean Sample Selection in Learning with Noisy Labels.
[arxiv] [code coming soon] [bib] [intro] [简介]
Yikai Wang*, Yanwei Fu*, Xinwei Sun†. Preprint, 2023. @article{wang2023knockoffs,
title={Knockoffs-SPR: Clean Sample Selection in Learning with Noisy Labels},
author={Wang, Yikai and Fu, Yanwei and Sun, Xinwei},
journal={arXiv preprint arXiv:2301.00545},
year={2023}
}

Scalable Penalized Regression for Noise Detection in Learning with Noisy Labels.
[arxiv] [paper] [code] [bib] [intro] [简介]
Yikai Wang, Xinwei Sun, Yanwei Fu†. CVPR, 2022. @inproceedings{wang2022scalable,
title={Scalable Penalized Regression for Noise Detection in Learning with Noisy Labels},
author={Wang, Yikai and Xinwei Sun and Fu, Yanwei},
booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2022}
}

How to Trust Unlabeled Data? Instance Credibility Inference for Few-Shot Learning.
[arxiv] [paper] [code] [bib] [intro]
Yikai Wang, Li Zhang, Yuan Yao†, Yanwei Fu†. TPAMI. Accepted in 2021.06. @article{wang2021trust,
title={How to Trust Unlabeled Data? Instance Credibility Inference for Few-Shot Learning},
author={Wang, Yikai and Zhang, Li and Yao, Yuan and Fu, Yanwei},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
year={2021},
doi={10.1109/TPAMI.2021.3086140}}
}

Instance Credibility Inference for Few-Shot Learning.
[arxiv] [paper] [code] [bib] [intro] [简介]
Yikai Wang, Chengming Xu, Chen Liu, Li Zhang, Yanwei Fu†. CVPR, 2020. @inproceedings{wang2020instance,
title={Instance Credibility Inference for Few-Shot Learning},
author={Wang, Yikai and Xu, Chengming and Liu, Chen and Zhang, Li and Fu, Yanwei},
booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2020}
}

An Embarrassingly Simple Baseline to One-Shot Learning.
[paper] [code] [bib]
Chen Liu, Chengming Xu, Yikai Wang, Li Zhang, Yanwei Fu†. CVPR Workshop, 2020. @InProceedings{Liu_2020_CVPR_Workshops,
author = {Liu, Chen and Xu, Chengming and Wang, Yikai and Zhang, Li and Fu, Yanwei},
title = {An Embarrassingly Simple Baseline to One-Shot Learning},
booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
year = {2020}
}

Generative Decoding

Entity-Level Text-Guided Image Manipulation.
[arxiv] [full-size PDF] [code] [bib] [intro] [demo: Youtube, Bilibili]
Yikai Wang*, Jianan Wang*, Guansong Lu, Hang Xu, Zhenguo Li, Wei Zhang, Yanwei Fu†. Preprint, 2023.
@article{wang2023entitylevel,
title={Entity-Level Text-Guided Image Manipulation},
author={Wang, Yikai and Wang, Jianan and Lu, Guansong and Xu, Hang and Li, Zhenguo and Zhang, Wei and Fu, Yanwei},
year={2023},
journal={arXiv preprint arXiv:2302.11383},
}

A Unified Prompt-Guided In-Context Inpainting Framework for Reference-based Image Manipulations.
[arxiv] [bib]
Chenjie Cao, Qiaole Dong, Yikai Wang, Yunuo Cai, Yanwei Fu†. Preprint, 2023.
@article{cao2023unified,
title={A Unified Prompt-Guided In-Context Inpainting Framework for Reference-based Image Manipulations},
author={Chenjie Cao and Qiaole Dong and Yikai Wang and Yunuo Cai and Yanwei Fu},
year={2023},
journal={arXiv preprint arXiv:2305.11577},
}

Joint fMRI Decoding and Encoding with Latent Embedding Alignment.
[arxiv] [bib]
Xuelin Qian*, Yikai Wang*, Yanwei Fu, Xinwei Sun, Jianfeng Feng, Xiangyang Xue. Preprint, 2023. Work in progress.
@article{qian2023semantic,
title={Joint fMRI Decoding and Encoding with Latent Embedding Alignment},
author={Xuelin Qian and Yikai Wang and Yanwei Fu and Xinwei Sun and Jianfeng Feng and Xiangyang Xue},
year={2023},
journal={arXiv preprint arXiv:2303.14730},
}

Miscellaneous

FFD Augmentor: Towards Few-Shot Oracle Character Recognition from Scratch.
[paper] [code] [bib]
Xinyi Zhao, Siyuan Liu, Yikai Wang, Yanwei Fu†. ACCV, 2022.
Note: This is the final project of undergraduate course NNDL 2022, for which I served as TA. @InProceedings{Zhao_2022_ACCV,
author = {Zhao, Xinyi and Liu, Siyuan and Wang, Yikai and Fu, Yanwei},
title = {FFD Augmentor: Towards Few-Shot Oracle Character Recognition from Scratch},
booktitle = {Proceedings of the Asian Conference on Computer Vision (ACCV)},
month = {December},
year = {2022},
pages = {1622-1639}
}

Service


Teaching


Network and Deep Learning,
TA, 2019-2022 Springs in School of Data Science, Fudan University.

Education


Ph.D. of Statistics (Machine Learning track),
2019.9 - Present, Fudan University, Shanghai, China.
Bachelor of Science in Mathmatics,
2015.9 - 2019.6, Fudan University, Shanghai, China.