Aug 24, 2024 · We propose the FedTSA for inter-client collaborative Test-Time Adaptation (TTA), targeting the deployment of pre-trained models under testing distribution ...
Duration: 1:59
Posted: Jul 11, 2024
Posted: Jul 11, 2024
In federated learning (FL), the not independently or identically distributed (non-IID) data partitioning impairs the performance of the global model, which is a ...
Sort by Relevance. Article/Chapter Results. Enabling Collaborative Test-Time Adaptation in Dynamic Environment via Federated Learning. Zhang, Jiayuan;Liu ...
Nov 5, 2024 · This paper explores test-time adaptation (TTA) in scenarios involving varying computational resources across multiple devices. To address this, ...
A curated list of awesome online test-time adaptation resources. Your contributions are always welcome!
引用次数 ; Enabling Collaborative Test-Time Adaptation in Dynamic Environment via Federated Learning. J Zhang, X Liu, Y Zhang, G Zhu, J Niu, S Tang. Proceedings ...
Test-time adaptation (TTA) is designed to enable existing models to quickly adapt to new target data without ac- cessing source domain data. As an important ...
Enabling Collaborative Test-Time Adaptation in Dynamic Environment via Federated Learning. 2024-08-25 | Conference paper. DOI: 10.1145/3637528.3671908.
Aug 24, 2024 · This study found that existing TTA methods consider only the largest logit as a pseudo-label and aim to minimize the entropy of test time predictions.