Federated Computing -- Survey on Building Blocks, Extensions and Systems

Federated Computing enables data privacy while supporting AI development: Rather than centralizing sensitive data in one location, Federated Computing allows organizations and individuals to keep their data locally while still participating in collaborative AI training and analysis, addressing both security concerns and regulatory requirements like data protection laws.

Data sovereignty is a core principle, not just a technical feature: The paper emphasizes that Federated Computing aligns with "user empowerment and data sovereignty," meaning individuals and organizations maintain control over their own data throughout the collaborative process—a key requirement for responsible AI practices.

The field lacks clear standards and definitions: The survey of over 150 papers reveals that Federated Computing systems "currently lack a clear definition and taxonomy," suggesting the field would benefit from standardized frameworks to ensure consistent implementation of privacy protections and ethical practices across different applications.

Federated Learning and Federated Analytics serve different purposes: The paper distinguishes between Federated Learning (training AI models collaboratively) and Federated Analytics (analyzing data collaboratively), noting "unique differences between those two"—an important distinction for organizations choosing which approach fits their responsible AI goals.

Federated Computing addresses real-world regulatory and security challenges: The approach directly responds to "data security breaches and regulatory hurdles" faced by centralized systems, making it practically relevant for organizations struggling to balance innovation with compliance and data protection obligations.