Target Score Identity
A mathematical identity constructed from an unnormalized probability density that expresses the score function as a self-consistent relation between model predictions and the density gradient, allowing diffusion models to be trained without external training data.
Timeline (last 90 days)
The core innovation of fixed-point score matching is replacing the denoising score identity in stochastic interpolation with a target score identity, substituting samples generated by the current model for the true target samples, thereby shifting the expectation from computation under the target distribution to computation under the model distribution.
The target score identity expresses the score function as a self-consistent relation between the current model prediction and the gradient of the unnormalized density, making it possible to construct a training objective without external training data.
All Facts (2)
The core innovation of fixed-point score matching is replacing the denoising score identity in stochastic interpolation with a target score identity, substituting samples generated by the current model for the true target samples, thereby shifting the expectation from computation under the target distribution to computation under the model distribution.
50%UnverifiedThe target score identity expresses the score function as a self-consistent relation between the current model prediction and the gradient of the unnormalized density, making it possible to construct a training objective without external training data.
50%