On the Security & Privacy in Federated Learning
This work addresses security and privacy issues in FL, which is critical for industries adopting this technology, but it is incremental as it synthesizes existing knowledge rather than introducing new methods.
The paper tackles the problem of security and privacy vulnerabilities in Federated Learning (FL) by reviewing state-of-the-art threats and defenses, resulting in the creation of a unifying taxonomy and threat model to guide future research.
Recent privacy awareness initiatives such as the EU General Data Protection Regulation subdued Machine Learning (ML) to privacy and security assessments. Federated Learning (FL) grants a privacy-driven, decentralized training scheme that improves ML models' security. The industry's fast-growing adaptation and security evaluations of FL technology exposed various vulnerabilities that threaten FL's confidentiality, integrity, or availability (CIA). This work assesses the CIA of FL by reviewing the state-of-the-art (SoTA) and creating a threat model that embraces the attack's surface, adversarial actors, capabilities, and goals. We propose the first unifying taxonomy for attacks and defenses and provide promising future research directions.