Informationally Compressive Anonymization: Non-Degrading Sensitive Input Protection for Privacy-Preserving Supervised Machine Learning

ArXi:2603.15842v1 Announce Type: cross Modern machine learning systems increasingly rely on sensitive data, creating significant privacy, security, and regulatory risks that existing privacy-preserving machine learning (ppML) techniques, such as Differential Privacy (DP) and Homomorphic Encryption (HE), address only at the cost of degraded performance, increased complexity, or prohibitive computational overhead. This paper