Survey on federated learning threats: Concepts, taxonomy on attacks and defences, experimental study and challenges

Note

Hey guys, this is my personal reading note. I am not sure there might be some mistakes in my understanding. Please feel free to correct me (hsiangjenli@gmail.com) if you find any. Thanks!

• Publish Year : 2023

• Authors : Rodríguez-Barroso, Jiménez-López, Luzón, Herrera, and Martínez-Cámara

Abstract

• federated learning is threatened by adversarial attacks against the integrity of the learning model and the privacy of data via a distributed approach to tackle local and global learning

• A taxonomy of

  1. adversarial attacks (e.g., ….)

  2. defence methods (e.g., ….)

that depict a general picture of this vulnerability of federated learning and how to overcome it

Contributions

Introduction

• Adversarial attacks

  • malicious manipulation of the training data [2]

Differnt types of Federated Learning

1. Horizontal Federated Learning (HFL) 共同的 Feature space

   • Each data source has the differnt feature space and sample space

   • HFL will only use the overlapped feature space to train the model

2. Vertical Federated Learning (VFL) 共同的 Sample space

3. Federated Transfer Learning (FTL)

Term Explanation

• Federated Learning aims at generating a collaboratively trained global learning model without sharing the data owned by the distributed data sources

Reference

[1] (1,2)

Nuria Rodríguez-Barroso, Daniel Jiménez-López, M Victoria Luzón, Francisco Herrera, and Eugenio Martínez-Cámara. Survey on federated learning threats: concepts, taxonomy on attacks and defences, experimental study and challenges. Information Fusion, 90:148–173, 2023.

[2]

Nilesh Dalvi, Pedro Domingos, Mausam, Sumit Sanghai, and Deepak Verma. Adversarial classification. In Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, 99–108. 2004.