Generic Folding Scheme Enables Efficient Non-Uniform Verifiable Computation
Protostar introduces a generic folding scheme for special-sound protocols, drastically reducing recursive overhead for complex, non-uniform verifiable computation.
ZKTorch: Efficient, Private ML Inference via Parallel Zero-Knowledge Proof Accumulation
ZKTorch enables private, verifiable ML inference by compiling models into basic blocks, leveraging parallel proof accumulation for efficiency.
Folding Schemes Enable Efficient Recursive Zero-Knowledge Arguments
A new cryptographic primitive, the folding scheme, dramatically reduces recursive proof overhead, unlocking practical incrementally verifiable computation.
ZKTorch: Efficiently Verifying ML Inference with Zero-Knowledge Proofs
ZKTorch introduces a parallel proof accumulation system for ML inference, fundamentally enhancing transparency while safeguarding proprietary model weights.
