Encrypted Computation - What if Decryption Wasn't Needed?

Explore the revolutionary field of encrypted computation in this 35-minute conference talk from GOTO Copenhagen 2024, where Principal Data Scientist Katharine Jarmul demonstrates how to perform calculations and operations on data without ever needing to decrypt it. Discover the core mathematical theory behind homomorphic encryption and secure multi-party computation, understanding why and how these cryptographic techniques work to solve data privacy challenges in machine learning and data science. Learn about the Paillier cryptosystem through practical demonstrations, including a live demo using Huggingface, and see how these technologies enable secure data processing while maintaining privacy. Examine real-world use cases for encrypted computation and gain insights into when and where these approaches provide significant advantages over traditional methods. Get practical guidance on open-source libraries and tools to start implementing encrypted computation solutions, including resources from TF Encrypted, Apple's Swift Homomorphic Encryption, Zama AI, and Carbyne Stack. Understand how these privacy-preserving technologies can help solve problems that are currently being addressed inefficiently or not at all, while building trust through transparency and understanding of the underlying cryptographic principles.