We propose {\it HumanDiffusion,} a diffusion model trained from humans' perceptual gradients to learn an acceptable range of data for humans (i.e., human-acceptable distribution). Conventional HumanGAN aims to model the human-acceptable distribution wider than the real-data distribution by training a neural network-based generator with human-based discriminators. However, HumanGAN training tends to converge in a meaningless distribution due to the gradient vanishing or mode collapse and requires careful heuristics. In contrast, our HumanDiffusion learns the human-acceptable distribution through Langevin dynamics based on gradients of human perceptual evaluations. Our training iterates a process to diffuse real data to cover a wider human-acceptable distribution and can avoid the issues in the HumanGAN training. The evaluation results demonstrate that our HumanDiffusion can successfully represent the human-acceptable distribution without any heuristics for the training.
@article{arxiv.2306.12169,
title = {HumanDiffusion: diffusion model using perceptual gradients},
author = {Yota Ueda and Shinnosuke Takamichi and Yuki Saito and Norihiro Takamune and Hiroshi Saruwatari},
journal= {arXiv preprint arXiv:2306.12169},
year = {2023}
}