HarmoFL
by med-air
[AAAI'22] HarmoFL: Harmonizing Local and Global Drifts in Federated Learning on Heterogeneous Medical Images
AI summary
Drift harmonizer
A framework to harmonize local and global drifts in federated learning on heterogeneous medical images
- stars
- 86
- forks
- 15
- watching
- 4
Similar projects
Found by comparing what the projects do, not just their names.
Feature normalization method
An approach to federated learning that addresses feature shift non-iid by normalizing local batch features before averaging models.
Data heterogeneity solver
A federated learning framework designed to mitigate data heterogeneity in distributed learning settings.
Shift robust FL framework
A framework for personalized federated learning that improves shift-robustness with minimal extra training overhead
DPFL Algorithm
This repository provides an implementation of a differentially private federated learning algorithm designed to improve the robustness and performance of federated machine learning systems.
Federated learner
Federated learning algorithm that adapts to non-IID data by decoupling and correcting for local drift
Drift detector
A Python library that detects drifts in machine learning systems
Federated Learning Library
An implementation of efficient federated learning algorithms for heterogeneous clients
Federated learning enhancer
Improves federated learning models by addressing data heterogeneity through distributional transformation
Federated Learning Framework
An implementation of a heterogenous federated learning framework using model distillation.
Federated Image Segmentation Framework
This project presents a framework for federated domain generalization in medical image segmentation using continuous frequency space and episodic learning.
Federated Learning framework
An implementation of heterogeneous federated learning with parallel edge and server computation
Federated distillation framework
A framework for achieving domain-aware knowledge distillation in federated learning environments.
Data processing framework
This project provides tools and frameworks to mitigate hallucinatory toxicity in visual instruction data, allowing researchers to fine-tune MLLM models on specific datasets.
Distributed Backpropagation Library
Enables backpropagation in distributed settings and facilitates model parallelism using differentiable communication between processes
Merkle DAG library
A Merkle DAG implementation for replicating immutable data in a peer-to-peer network