Software/algorithm Side: Data Encryption

In the digital age, where data security and privacy are critical, the need to protect user data within decentralized inference systems like Nesa cannot be overstated. As decentralized systems distribute computational tasks across various nodes, each potentially operated by different entities, the risk of exposing sensitive user data during the inference process intensifies. Therefore, ensuring user data's confidentiality and integrity becomes a cornerstone of trust and reliability in such systems. This section discusses our vision and implementation of homomorphic encryption and split learning to achieve this goal.