Privacy Preserving Inference

Privacy-preserving inference aims to perform computations on sensitive data without revealing the data itself, primarily focusing on machine learning models. Current research emphasizes efficient techniques for handling non-linear operations within various architectures, including large language models (LLMs), graph neural networks (GNNs), and Kolmogorov-Arnold Networks (KANs), often employing methods like homomorphic encryption, polynomial approximations, and secure multi-party computation. This field is crucial for enabling the deployment of powerful machine learning models in privacy-sensitive applications, such as healthcare and finance, while mitigating data breaches and enhancing user trust.