About la$LA
How does Lagrange work?
Lagrange relies on zero-knowledge proofs, a cryptographic technique that allows one party to prove the validity of a computation without revealing its underlying data. It combines a ZK proving technique called DeepProve with its ZK Coprocessor and Prover Network. This combination enables complex tasks to be computed while providing on-chain verifiability that they were executed as intended. Let’s look at each of these components in detail below.
DeepProve
Lagrange’s zero-knowledge machine learning (zkML) framework allows developers to cryptographically prove the correctness of AI model inferences without revealing the underlying model parameters or data. This can be used by projects to protect their proprietary data while proving the authenticity of the output.
By utilizing advanced cryptographic techniques, DeepProve achieves proving times that are significantly faster than previous zkML solutions. For example, it can prove the inference of a neural network up to 158 times faster than leading alternatives.
