- Published on
2026년 프라이버시 보존 머신러닝(PPML) 지형 - Federated Learning(Flower 1.13, NVIDIA FLARE 2.5, TensorFlow Federated, PySyft 0.9), Differential Privacy(Opacus 1.5, OpenDP 0.10, Google DP Library), Homomorphic Encryption(Zama Concrete, Microsoft SEAL, OpenFHE), Secure MPC(CrypTen, MP-SPDZ, EzPC), TEE(Intel SGX, Apple PCC, AWS Nitro Enclaves)까지. Google Gboard, Apple Private Cloud Compute, BraTS 의료 컨소시엄, 한국/일본 금융권 실제 배포 사례와 GDPR/HIPAA/EU AI Act/K-PIPA/APPI 규제 대응을 다룹니다.