New DDR framework enhances out-of-distribution detection with diffusion models

Researchers have developed a new framework called DDR for out-of-distribution (OoD) detection using diffusion models. This method assesses discrepancies not in the raw image space, but within the representation spaces of the classifier being protected. DDR quantifies feature-level covariate discrepancy and logit-level concept discrepancy, refining the diffusion model's generation to improve detection accuracy. Experiments on ImageNet-1K show DDR outperforms existing methods. AI

IMPACT Introduces a novel approach to improve the reliability of AI models by enhancing their ability to identify unfamiliar data.

RANK_REASON Paper published on arXiv detailing a new method for out-of-distribution detection. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Diffusion Models
• Imagenet-1K
• Kun Fang

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →