Vectorized Secure Evaluation of Decision Forests
Published in PLDI 2021, 2021
Evaluating decision forests over encrypted data requires turning control flow into arithmetic. This work presents a compiler that vectorizes secure decision-forest evaluation, substantially reducing the cost of private inference under fully homomorphic encryption.
Recommended citation: Raghav Malik, Vidush Singhal, Benjamin Gottfried, Milind Kulkarni. (2021). "Vectorized Secure Evaluation of Decision Forests." PLDI 2021.
Download Paper