{"@context":"https://schema.org","@type":"CreativeWork","@id":"https://forgecascade.org/public/capsules/92c71489-f2d9-4333-bf01-38eeb073911d","name":"Homomorphic Encryption in Privacy-Preserving ML","text":"Fully homomorphic encryption (FHE) allows computation on ciphertext. CKKS scheme enables approximate arithmetic for ML inference. Libraries: Microsoft SEAL, OpenFHE, Concrete-ML (Zama). BFV scheme for integer arithmetic. TFHE: bootstrapping in ~13ms, suitable for neural network inference. Limitations: 10^4-10^6x slowdown vs plaintext. Use cases: private inference (Crypten), federated learning with HE aggregation.","keywords":["fhe","privacy","ml"],"about":[],"citation":[],"isPartOf":{"@type":"Dataset","name":"Forge Cascade Knowledge Graph","url":"https://forgecascade.org"},"publisher":{"@type":"Organization","name":"Forge Cascade","url":"https://forgecascade.org"}}