paper-with-me

홈 › Papers

Generative Data Mining with Longtail-Guided Diffusion

2025-02-04 · David S. Hayden, Mao Ye, Timur Garipov, Gregory P. Meyer, Carl Vondrick, Zhao Chen, Yuning Chai, Eric Wolff, Siddhartha S. Srinivasa

It is difficult to anticipate the myriad challenges that a predictive model will encounter once deployed. Common practice entails a reactive, cyclical approach: model deployment, data mining, and retraining. We instead develop a proactive longtail discovery process by imagining additional data during training. In particular, we develop general model-based longtail signals, including a differentiable, single forward pass formulation of epistemic uncertainty that does not impact model parameters or predictive performance but can flag rare or hard inputs. We leverage these signals as guidance to generate additional training data from a latent diffusion model in a process we call Longtail Guidance (LTG). Crucially, we can perform LTG without retraining the diffusion model or the predictive model, and we do not need to expose the predictive model to intermediate diffusion states. Data generated by LTG exhibit semantically meaningful variation, yield significant generalization improvements on image classification benchmarks, and can be analyzed to proactively discover, explain, and address conceptual gaps in a predictive model.

📄 PDF Abstract BibTeX arXiv:2502.01980

Code (0)

등록된 구현이 없습니다.

Tasks

image-classificationImage Classification

Methods 이 논문이 사용한 방법론

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…
Latent Diffusion Model Diffusion models applied to latent spaces, which are normally built with (Variational) Autoencoders.

Similar Papers 제목 키워드 기반

OpenLongTail: Generative Scaling of Long-Tail Driving Data

2026-07-10 · Lulin Liu, Nuo Chen, Yan Wang, Bangya Liu 외 arxiv

Scaling robust driving policies is fundamentally bottlenecked by the scarcity of edge cases in curated datasets. While the real world continuously captures these critical events, such long-tail events remain underutilize…

Autonomous Driving

Circuit-Aware Reward Training: A Mechanistic Framework for Longtail Robustness in RLHF

2025-09-29 · Jing Liu arxiv

Reinforcement Learning from Human Feedback (RLHF) reward models exhibit systematic failures on longtail distributions, leading to reward hacking and misalignment. We propose a mechanistic interpretability framework that …

Reinforcement LearningData Augmentation

Reasoning models do not yet follow their reasoning in autonomous driving: The KITScenes LongTail Dataset

2026-03-24 · Royden Wagner, Omer Sahin Tas, Jaime Villa, Felix Hauser 외 arxiv

Handling rare events is the central open challenge in autonomous driving. Reasoning models, which generate explicit chains of reasoning before acting, promise to generalize to such events. Here we show that these models …

Instruction Following

A Survey of Multimodal-Guided Image Editing with Text-to-Image Diffusion Models

2024-06-20 · Xincheng Shuai, Henghui Ding, Xingjun Ma, RongCheng Tu 외

Image editing aims to edit the given synthetic or real image to meet the specific requirements from users. It is widely studied in recent years as a promising and challenging field of Artificial Intelligence Generative C…

Video Editing

Manifold-Constrained Adversarial Training for Long-Tailed Robustness via Geometric Alignment

2026-05-04 · Guanmeng Xian, Ning Yang, Philip S. Yu arxiv

Adversarial training is effective on balanced datasets, but its robustness degrades under longtailed class distributions, where tail classes suffer high robust error and unstable decision boundaries. We propose Manifold-…

Adversarial Robustness