The book presents theory and algorithms for secure networked inference in the presence of Byzantines. It derives fundamental limits of networked inference in the presence of Byzantine data and designs robust strategies to ensure reliable performance for several practical network architectures. In particular, it addresses inference (or learning) processes such as detection, estimation or classification, and parallel, hierarchical, and fully decentralized (peer-to-peer) system architectures. Furthermore, it discusses a number of new directions and heuristics to tackle the problem of design complexity in these practical network architectures for inference.

The book presents theory and algorithms for secure networked inference in the presence of Byzantines. It derives fundamental limits of networked inference in the presence of Byzantine data and designs robust strategies to ensure reliable performance for several practical network architectures. In particular, it addresses inference (or learning) processes such as detection, estimation or classification, and parallel, hierarchical, and fully decentralized (peer-to-peer) system architectures. Furthermore, it discusses a number of new directions and heuristics to tackle the problem of design complexity in these practical network architectures for inference.

The book presents theory and algorithms for secure networked inference in the presence of Byzantines. It derives fundamental limits of networked inference in the presence of Byzantine data and designs robust strategies to ensure reliable performance for several practical network architectures. In particular, it addresses inference (or learning) processes such as detection, estimation or classification, and parallel, hierarchical, and fully decentralized (peer-to-peer) system architectures. Furthermore, it discusses a number of new directions and heuristics to tackle the problem of design complexity in these practical network architectures for inference.