Kai Yao 姚 凯

Kai dot Yao at liverpool dot ac dot uk

I'm a researcher in Ant Group working on privacy-preserving large fondationa models. Meanwhile, I am a PostDoc in University of Zhejiang advised by Prof. Jianke Zhu and CTO Wei Wang. My research background lies in computer vision and transfer learning.

English CV  /  Chinese CV  /  Google Scholar  /  Github

• DEC 2023 - Our papers accepted in AAAI24.
• Aug 2023 - Our papers accepted as Oral Presentation in ACM CIKM2023.
• May 2023 - Our two journal papers accepted in IEEE TETCI and IJB.
• show more
Nov 2022 - Our conference paper accepted in AAAI'23. We propose a novel augmentation for medical image domain generalization. The code released at here.
Jul 2022 - Our conference paper accepted in ECCV'22. We propose a novel hybrid vision-transformer-based GAN for image outpainting. The code released at here.
Jun 2022 - Our journal paper accepted in IEEE-JBHI (IF: 7.0).
Dec 2021 - Our team PremiLab wins 5th in the Challenge CrossMoDa 2021 (MICCAI workshop).
Dec 2021 - Our journal paper accepted in International Journal of Bioprinting (IF: 7.4).
Nov 2021 - Our journal paper accepted in Cognitive Computation (IF: 4.9).
Dec 2019 - I started my PhD course at UoL PremiLab Lab with full scholarship.

Sort by publication time. First author papers are highlighted.

A novel strategy for training-free test-time adaptation.

A novel episteme-based framework for domain adaptive semantic segmentation.

A review paper summarize past works and reveal the current challenging towards bioprinting.

In this study, we proposed a novel keypoint-aware method to to simultaneously detect, segment, and classify nuclei from Haematoxylin and Eosin (H&E) stained histopathology data.

We propose a novel strategy to alleviate the domain gap imbalance considering the characteristics of medical images, namely Global-Local Union Alignment.

This paper overviews the status and progress of ML applications for skin disease recognition.

We design a novel augmentation for medical image domain generalization and theoretically prove that our proposed augmentation can lead to an upper bound of the generalization risk on the unseen target domain.

we propose a novel hybrid vision-transformer-based encoder-decoder framework, named Query Outpainting TRansformer (QueryOTR), for extrapolating visual context all-side around a given image.

We proposed a novel GAN for diverse style transfer, and a 3D Dual Attention Residual U-Net for robust and reliable cross-modality semantic segmentation.

This paper overviews the status and progress of ML applications from several aspects: parameter optimization in the fabrication model, in situ monitoring and control, scaffold performance evaluation, and material design.

Taking advantages of AD-GAN, in this paper, we study with cell-scaffold interaction.

We developed herein an end-to-end model called Aligned Disentangled Generative Adversarial Network (AD-GAN) for 3D unsupervised nuclei segmentation of CLSM images.

The purpose of this work is to build and evaluate the performance of ultrathin scaffolds with an electrohydrodynamic jet (EHDJ) printing method for RPE cell culture.

In this paper, we introduce a new benchmark nuclei segmentation dataset termed as Scaffold-A549 for 3D cell culture on bio-scaffold.

In this paper, I mainly contribute to the visualization.

In this paper, we propose to integrate a kernelized Min-Max objective in the DNN training in order to explicitly enforce both kernelized within-class compactness and between-class margin.

In this paper, we propose to integrate a kernelized Min-Max objective in the DNN training in order to explicitly enforce both kernelized within-class compactness and between-class margin.

Thanks to the source code from this guy.