[
    {
        "id": "osp-16522",
        "type": "article-journal",
        "title": "Enabling Preference-driven Unlearning in Few-step Distilled Text-to-Image Diffusion Models",
        "author": [
            {
                "family": "Patel",
                "given": "Gaurav"
            },
            {
                "family": "Fang",
                "given": "Jun"
            },
            {
                "family": "Steeg",
                "given": "Greg Ver"
            },
            {
                "family": "Qiu",
                "given": "Qiang"
            },
            {
                "family": "Sripada",
                "given": "Sravan"
            }
        ],
        "URL": "https://omanscience.com/ar/articles/enabling-preference-driven-unlearning-in-few-step-distilled-text-to-image-diffusion-models",
        "language": "en",
        "issued": {
            "date-parts": [
                [
                    2026
                ]
            ]
        },
        "abstract": "Text-to-image diffusion models are increasingly distilled into few-step variants and being deployed to enable fast inference. However, their ability to generate harmful or undesired content poses significant safety risks. Data-driven unlearning methods suppress targeted generations by fine-tuning model weights using specialized unlearning objectives. Crucially, these objectives implicitly rely on multi-step denoising dynamics, an assumption that breaks down for few-step distilled (FSD) models, resulting in ineffective forgetting. Furthermore, performing unlearning on the non-distilled base model and subsequently re-distilling it to obtain an unlearned FSD model incurs substantial computational and time overhead, making it impractical in many settings. Hence, we address this limitation with a preference-driven unlearning framework that revisits Direct Preference Optimization (DPO) for diffusion models. We show that standard DPO and its unlearning derivatives, formulated around noise-prediction error, transfer poorly to FSD models due to their altered generation dynamics. To overcome this, we introduce a modified preference optimization formulation explicitly aligned with the few-step generation properties, enabling direct concept removal in FSD models while preserving few-step efficiency and maintaining strong retention of desirable (non-targeted) capabilities. We evaluate our framework primarily on identity and NSFW (nudity) removal tasks and also extend our method to object-level unlearning. Extensive experiments demonstrate consistent and effective forgetting, and strong retention performance, establishing our method as a practical and principled solution for unlearning in FSD models."
    }
]