{"slug": "probing-diffusion-denoising-dynamics-for-contrastive-representation-learning", "title": "Probing Diffusion Denoising Dynamics for Contrastive Representation Learning", "summary": "Researchers introduced D³CL, a parameter-efficient framework that adapts pretrained text-to-image diffusion models for discriminative representation learning by coupling contrastive objectives with denoising reconstruction. On ImageNet-1K, D³CL achieved 80.1% linear-probing accuracy and an FID of 5.56 for unconditional generation, demonstrating that noise-level contrastive learning can enhance discriminative tasks while preserving generative performance.", "body_md": "arXiv:2607.09067v1 Announce Type: new\nAbstract: Text-to-image diffusion models exhibit unprecedented generative capability and contain rich intermediate representations that can be useful for discriminative vision tasks. Motivated by this observation, we study a focused question: how can the denoising dynamics of a pretrained diffusion model be adapted to support discriminative representation learning while preserving its generative behavior under parameter-efficient updates? We present D$^3$CL as an investigation of this question. Our key observation is that noisy latents at different diffusion timesteps can be interpreted as stochastic views of the same underlying image, enabling a contrastive objective to be coupled with the standard denoising reconstruction loss. This formulation provides a simple way to probe the interaction between generative denoising and discriminative representation learning without training from scratch. To keep the adaptation lightweight, we apply LoRA updates to a pretrained Stable Diffusion backbone while freezing the original model parameters. D$^3$CL provides strong empirical evidence that reconstruction and noise-level contrastive objectives can be complementary: on ImageNet-1K, it obtains 80.1% linear-probing accuracy and an FID of 5.56 for $256 \\times 256$ unconditional generation. Additional ablations on the design space suggest that the usefulness of diffusion features depends on where and how denoising states are sampled. These results establish D$^3$CL as a parameter-efficient adaptation framework for pretrained diffusion models, showing that noise-level contrastive learning can structure denoising representations for discriminative tasks while maintaining generative performance.", "url": "https://wpnews.pro/news/probing-diffusion-denoising-dynamics-for-contrastive-representation-learning", "canonical_source": "https://arxiv.org/abs/2607.09067", "published_at": "2026-07-13 04:00:00+00:00", "updated_at": "2026-07-13 04:25:28.327620+00:00", "lang": "en", "topics": ["machine-learning", "computer-vision", "generative-ai", "artificial-intelligence"], "entities": ["D³CL", "Stable Diffusion", "ImageNet-1K"], "alternates": {"html": "https://wpnews.pro/news/probing-diffusion-denoising-dynamics-for-contrastive-representation-learning", "markdown": "https://wpnews.pro/news/probing-diffusion-denoising-dynamics-for-contrastive-representation-learning.md", "text": "https://wpnews.pro/news/probing-diffusion-denoising-dynamics-for-contrastive-representation-learning.txt", "jsonld": "https://wpnews.pro/news/probing-diffusion-denoising-dynamics-for-contrastive-representation-learning.jsonld"}}