<ul><li>Fine-tuning text-to-image diffusion models with human feedback is an effective method for aligning model behavior with human intentions.</li><li>A novel automated data filtering algorithm called FiFA is proposed to enhance the fine-tuning of diffusion models using human feedback datasets with preference optimization.</li><li>FiFA selects data based on preference margin, text quality, and text diversity, ensuring informative samples and prevention of harmful content.</li><li>Experimental results show that FiFA significantly improves training stability and achieves better performance with reduced data usage.</li></ul>

Automated Filtering of Human Feedback Data for Aligning Text-to-Image Diffusion Models

Discover more