Comparison of Fully Homomorphic Encryption and Garbled Circuit Techniques in Privacy-Preserving Machine Learning Inference

The project is a research artifact or academic study comparing two established cryptographic primitives (FHE and GC) in the context of ML inference. With 0 stars and 3 forks after six months, it lacks community traction and functional utility as a standalone tool. Its value lies in the comparative data rather than the code itself. In the competitive landscape, specialized companies like Zama (FHE), Duality Technologies, and Inpher provide production-grade versions of these technologies with significantly deeper moats and optimization. Frontier labs like Google are already active in the FHE space (e.g., Google's 'fully-homomorphic-encryption' library), making this specific comparison easily reproducible and likely to be superseded by more advanced hardware-accelerated implementations or TEE-based (Trusted Execution Environment) solutions which are currently more performant for ML tasks. The displacement horizon is short because the field of PPML is evolving rapidly toward hybrid approaches (like Cheetah or Gazelle) which combine FHE and GC, rendering a simple binary comparison obsolete.