FedCorr: Multi-Stage Federated Learning for Label Noise Correction

2022 IEEE/CVF Conference on Computer Vision and Patter Recognition

Few-shot Image Generation via Adaptation-Aware Kernel Modulation

Thirty-Sixth Conference on Neural Information Processing Systems

Exact 3D Path Generation via 3D Cam-Linkage Mechanisms

ACM Transactions on Graphics

SUTD Author: Xu, Jingyi; Chen, Zihan; Quek, Tony Q. S.; Chong, Kai Fong Ernest

SUTD Author: Yunqing Zhao, Keshigeyan Chandrasegaran, Milad Abdollahzadeh, Ngai-Man Cheung

SUTD Author: Cheng, Yingjie; Song, Peng; Lu, Yukun; Chew, Wen Jie Jeremy

“This work presents a general multi-stage federated learning framework that tackles data heterogeneity, with respect to both local label quality and local data statistics. Without making any assumptions about the local noise model, this framework demonstrates robustness and outperformance over diverse heterogeneous and noisy settings, in a privacy-preserving manner.” --- Tony Quek

“In this work, we study few-shot image generation (FSIG), which aims to learn to generate new and diverse samples given an extremely limited number of samples from a domain, e.g., 10 training samples. We propose Adaptation-Aware kernel Modulation (AdAM) to address general FSIG of different source-target domain proximity. Extensive experimental results show that the proposed method consistently achieves SOTA performance across source/target domains of different proximity, including challenging setups when source and target domains are more apart.”  ---  Cheung Ngai-Man

" In this work, we introduce a new 3D cam-linkage mechanism for exact 3D path generation, which only consists of two 3D cams and five links. We propose an optimization-based approach to designing a 3D cam-linkage mechanism for exactly generating a continuous 3D path with arbitrary shape.” ---  Song Peng