Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models
This addresses safety and bias issues in widely used image generation models, which is critical for real-world applications, though it is an incremental improvement focused on mitigating existing model limitations.
The paper tackles the problem of inappropriate and biased image generation in text-conditioned diffusion models by introducing Safe Latent Diffusion (SLD), which removes or suppresses such content during the diffusion process without additional training, maintaining image quality and text alignment as demonstrated on the I2P test bed.
Text-conditioned image generation models have recently achieved astonishing results in image quality and text alignment and are consequently employed in a fast-growing number of applications. Since they are highly data-driven, relying on billion-sized datasets randomly scraped from the internet, they also suffer, as we demonstrate, from degenerated and biased human behavior. In turn, they may even reinforce such biases. To help combat these undesired side effects, we present safe latent diffusion (SLD). Specifically, to measure the inappropriate degeneration due to unfiltered and imbalanced training sets, we establish a novel image generation test bed-inappropriate image prompts (I2P)-containing dedicated, real-world image-to-text prompts covering concepts such as nudity and violence. As our exhaustive empirical evaluation demonstrates, the introduced SLD removes and suppresses inappropriate image parts during the diffusion process, with no additional training required and no adverse effect on overall image quality or text alignment.