Saved in:
Bibliographic Details
Main Authors: Salami, Riccardo, Buzzega, Pietro, Mosconi, Matteo, Verasani, Mattia, Calderara, Simone
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.02447
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910965426028544
author Salami, Riccardo
Buzzega, Pietro
Mosconi, Matteo
Verasani, Mattia
Calderara, Simone
author_facet Salami, Riccardo
Buzzega, Pietro
Mosconi, Matteo
Verasani, Mattia
Calderara, Simone
contents Federated Learning (FL) aims at unburdening the training of deep models by distributing computation across multiple devices (clients) while safeguarding data privacy. On top of that, Federated Continual Learning (FCL) also accounts for data distribution evolving over time, mirroring the dynamic nature of real-world environments. While previous studies have identified Catastrophic Forgetting and Client Drift as primary causes of performance degradation in FCL, we shed light on the importance of Incremental Bias and Federated Bias, which cause models to prioritize classes that are recently introduced or locally predominant, respectively. Our proposal constrains both biases in the last layer by efficiently finetuning a pre-trained backbone using learnable prompts, resulting in clients that produce less biased representations and more biased classifiers. Therefore, instead of solely relying on parameter aggregation, we leverage generative prototypes to effectively balance the predictions of the global model. Our method significantly improves the current State Of The Art, providing an average increase of +7.8% in accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2406_02447
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Federated Class-Incremental Learning with Hierarchical Generative Prototypes
Salami, Riccardo
Buzzega, Pietro
Mosconi, Matteo
Verasani, Mattia
Calderara, Simone
Machine Learning
Federated Learning (FL) aims at unburdening the training of deep models by distributing computation across multiple devices (clients) while safeguarding data privacy. On top of that, Federated Continual Learning (FCL) also accounts for data distribution evolving over time, mirroring the dynamic nature of real-world environments. While previous studies have identified Catastrophic Forgetting and Client Drift as primary causes of performance degradation in FCL, we shed light on the importance of Incremental Bias and Federated Bias, which cause models to prioritize classes that are recently introduced or locally predominant, respectively. Our proposal constrains both biases in the last layer by efficiently finetuning a pre-trained backbone using learnable prompts, resulting in clients that produce less biased representations and more biased classifiers. Therefore, instead of solely relying on parameter aggregation, we leverage generative prototypes to effectively balance the predictions of the global model. Our method significantly improves the current State Of The Art, providing an average increase of +7.8% in accuracy.
title Federated Class-Incremental Learning with Hierarchical Generative Prototypes
topic Machine Learning
url https://arxiv.org/abs/2406.02447