Nec Corporation

Zeus Sw Defender, LLC

Kyungtae Kim

Haifeng Chen

LuAn Tang

Junghwan Rhee

Xiao Yu

Chung Hwan Kim

Gyungho Lee

Changwoo Pyo

Hyungyu Lee

A computer-implemented method for efficient and scalable enclave protection for machine learning (ML) programs includes tailoring at least one ML program to generate at least one tailored ML program for execution within at least one enclave, and executing the at least one tailored ML program within the at least one enclave.

Efficient and scalable enclave protection for machine learning programs

An apparatus, method, and computer program product are provided for encrypting a function symbol with relocation. The apparatus includes a compiler module, a static linker module, and an encryptor module. The compiler module inserts sequences of instructions to decrypt function symbols to be randomized at runtime before indirect function calls. The compiler module inserts an instruction sequence at compile time to encrypt an operand register that receives a local function symbol in position-independent code (PIC), where a call or store instruction uses the register as an operand. The static linker module inserts an encoding section at link time. The encoding section includes two columns representing the sizes of function symbols in bits or bytes and the locations storing the function symbols to be encrypted at runtime. The encryptor module encrypts at runtime the function symbols whose sizes and stored memory locations are identified in the encoding section.

Systems and methods for function pointer protection by function symbol encryption

Growing community of inventors

West Lafayette, IN, United States of America

Kyungtae Kim