Heterogeneous Label

Heterogeneous label problems in federated learning arise when participating clients possess datasets with differing label distributions or annotation types, hindering the training of a robust global model. Current research focuses on mitigating this heterogeneity through techniques like contrastive representation learning, model concatenation instead of averaging, and personalized federated learning approaches incorporating attention mechanisms or cooperation between online and offline models. These advancements aim to improve the accuracy and robustness of federated learning models, particularly in applications like medical imaging where data scarcity and variability across institutions are common. The ultimate goal is to enable effective collaborative model training while preserving data privacy and addressing the challenges posed by inconsistent labeling practices.

Papers

April 27, 2024

FedCRL: Personalized Federated Learning with Contrastive Shared Representations for Label Heterogeneity in Non-IID Data
Chenghao Huang, Xiaolu Chen, Yanru Zhang, Hao Wang
Raw Data Contrastive Loss Personalized Federated Learning Contrastive Representation Learning Contrastive Representation Federated Contrastive Learning Heterogeneous Label

April 4, 2024

Approximate Gradient Coding for Privacy-Flexible Federated Learning with Non-IID Data
Okko Makkonen, Sampo Niemelä, Camilla Hollanti, Serge Kas Hanna
Federated Learning Raw Data Gradient Coding Privacy Framework Anonymised Data Heterogeneous Label

December 21, 2023

Fed-CO2: Cooperation of Online and Offline Models for Severe Data Heterogeneity in Federated Learning
Zhongyi Cai, Ye Shi, Wei Huang, Jingya Wang
Knowledge Transfer Data Heterogeneity Online Service Label Distribution Offline Model Heterogeneous Label

December 11, 2023

Exploiting Label Skews in Federated Learning with Model Concatenation
Yiqun Diao, Qinbin Li, Bingsheng He
Label Distribution Heterogeneous Label

July 31, 2023

Federated Learning for Data and Model Heterogeneity in Medical Imaging
Hussain Ahmad Madni, Rao Muhammad Umer, Gian Luca Foresti
Knowledge Distillation Raw Data Medical Imaging Data Heterogeneity Model Heterogeneity Heterogeneous Label

April 12, 2023

Unifying and Personalizing Weakly-supervised Federated Medical Image Segmentation via Adaptive Representation and Aggregation
Li Lin, Jiewei Wu, Yixiang Liu, Kenneth K. Y. Wong, Xiaoying Tang
Segmentation Performance Weak Supervision Medical Image Segmentation Task Data Aggregation Anti Unification Adaptive Representation Federated Medical Image Segmentation Heterogeneous Label

April 2, 2023

Personalized Federated Learning with Local Attention
Sicong Liang, Junchao Tian, Shujun Yang, Yu Zhang
Personalized Federated Learning Local Context pFL Method Heterogeneous Label Personalized Attention

January 13, 2023

Contrast with Major Classifier Vectors for Federated Medical Relation Extraction with Heterogeneous Label Distribution
Chunhui Du, Hao He, Yaohui Jin
Relation Extraction Label Distribution Biomedical Relation Extraction Heterogeneous Label Probability Vector

November 14, 2022

Multi-center anatomical segmentation with heterogeneous labels via landmark-based models
Nicolás Gaggion, Maria Vakalopoulou, Diego H. Milone, Enzo Ferrante
Pixel Level Landmark Detection Anatomical Segmentation Multi Center Heterogeneous Label Multiclass Segmentation

Heterogeneous Label

Papers

FedCRL: Personalized Federated Learning with Contrastive Shared Representations for Label Heterogeneity in Non-IID Data

Approximate Gradient Coding for Privacy-Flexible Federated Learning with Non-IID Data

Fed-CO2: Cooperation of Online and Offline Models for Severe Data Heterogeneity in Federated Learning

Exploiting Label Skews in Federated Learning with Model Concatenation

Federated Learning for Data and Model Heterogeneity in Medical Imaging

Unifying and Personalizing Weakly-supervised Federated Medical Image Segmentation via Adaptive Representation and Aggregation

Personalized Federated Learning with Local Attention

Contrast with Major Classifier Vectors for Federated Medical Relation Extraction with Heterogeneous Label Distribution

Multi-center anatomical segmentation with heterogeneous labels via landmark-based models