Aligning Diffusion Models by Optimizing Human Utility
This work addresses the challenge of aligning AI-generated images with human preferences for users of text-to-image models, offering a more efficient alternative to costly pairwise preference data, though it is incremental as it builds on existing alignment techniques.
The paper tackles the problem of aligning text-to-image diffusion models with human preferences by introducing Diffusion-KTO, which maximizes expected human utility using per-image binary feedback like likes or dislikes, resulting in superior performance over existing methods such as supervised fine-tuning and Diffusion-DPO in human judgment and metrics like PickScore and ImageReward.
We present Diffusion-KTO, a novel approach for aligning text-to-image diffusion models by formulating the alignment objective as the maximization of expected human utility. Since this objective applies to each generation independently, Diffusion-KTO does not require collecting costly pairwise preference data nor training a complex reward model. Instead, our objective requires simple per-image binary feedback signals, e.g. likes or dislikes, which are abundantly available. After fine-tuning using Diffusion-KTO, text-to-image diffusion models exhibit superior performance compared to existing techniques, including supervised fine-tuning and Diffusion-DPO, both in terms of human judgment and automatic evaluation metrics such as PickScore and ImageReward. Overall, Diffusion-KTO unlocks the potential of leveraging readily available per-image binary signals and broadens the applicability of aligning text-to-image diffusion models with human preferences.