Abstract
It is well known that training a denoising score-based diffusion modelsrequires tens of thousands of epochs and a substantial number of image data totrain the model. In this paper, we propose to increase the efficiency intraining score-based diffusion models. Our method allows us to decrease thenumber of epochs needed to train the diffusion model. We accomplish this bysolving the log-density Fokker-Planck (FP) Equation numerically to compute thescore \textit{before} training. The pre-computed score is embedded into theimage to encourage faster training under slice Wasserstein distance.Consequently, it also allows us to decrease the number of images we need totrain the neural network to learn an accurate score. We demonstrate through ournumerical experiments the improved performance of our proposed method comparedto standard score-based diffusion models. Our proposed method achieves asimilar quality to the standard method meaningfully faster.