中文
相关论文

相关论文: FlowErase-RL: Rethinking Concept Erasure as Reward…

200 篇论文

Erasing harmful or proprietary concepts from powerful text to image generators is an emerging safety requirement, yet current "concept erasure" techniques either collapse image quality, rely on brittle adversarial losses, or demand…

机器学习 · 计算机科学 2025-11-11 Abhiram Kusumba , Maitreya Patel , Kyle Min , Changhoon Kim , Chitta Baral , Yezhou Yang

Recent advancements in flow-matching have enabled high-quality text-to-image generation. However, the deterministic nature of flow-matching models makes them poorly suited for reinforcement learning, a key tool for improving image quality…

计算机视觉与模式识别 · 计算机科学 2025-10-06 Benjamin Yu , Jackie Liu , Justin Cui

We propose Flow-GRPO, the first method to integrate online policy gradient reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary…

计算机视觉与模式识别 · 计算机科学 2025-10-28 Jie Liu , Gongye Liu , Jiajun Liang , Yangguang Li , Jiaheng Liu , Xintao Wang , Pengfei Wan , Di Zhang , Wanli Ouyang

Recent flow matching models for text-to-image generation have achieved remarkable quality, yet their integration with reinforcement learning for human preference alignment remains suboptimal, hindering fine-grained reward-based…

计算机视觉与模式识别 · 计算机科学 2025-10-16 Xiaoxuan He , Siming Fu , Yuke Zhao , Wanli Li , Jian Yang , Dacheng Yin , Fengyun Rao , Bo Zhang

Concept Erasure, which aims to prevent pretrained text-to-image models from generating content associated with semantic-harmful concepts (i.e., target concepts), is getting increased attention. State-of-the-art methods formulate this task…

计算机视觉与模式识别 · 计算机科学 2025-08-07 Hongxu Chen , Zhen Wang , Taoran Mei , Lin Li , Bowei Zhu , Runshi Li , Long Chen

Generative speech enhancement offers a promising alternative to traditional discriminative methods by modeling the distribution of clean speech conditioned on noisy inputs. Post-training alignment via reinforcement learning (RL) effectively…

音频与语音处理 · 电气工程与系统科学 2026-01-26 Haoxu Wang , Biao Tian , Yiheng Jiang , Zexu Pan , Shengkui Zhao , Bin Ma , Daren Chen , Xiangang Li

Concept erasure techniques have been widely deployed in T2I diffusion models to prevent inappropriate content generation for safety and copyright considerations. However, as models evolve to next-generation architectures like Flux,…

机器学习 · 计算机科学 2025-10-07 Daiheng Gao , Nanxiang Jiang , Andi Zhang , Shilin Lu , Yufei Tang , Wenbo Zhou , Weiming Zhang , Zhaoxin Fan

We propose FlowRL: matching the full reward distribution via flow balancing instead of maximizing rewards in large language model (LLM) reinforcement learning (RL). Recent advanced reasoning models adopt reward-maximizing methods (\eg, PPO…

Recent GRPO-based approaches built on flow matching models have shown remarkable improvements in human preference alignment for text-to-image generation. Nevertheless, they still suffer from the sparse reward problem: the terminal reward of…

计算机视觉与模式识别 · 计算机科学 2026-02-26 Haoyou Deng , Keyu Yan , Chaojie Mao , Xiang Wang , Yu Liu , Changxin Gao , Nong Sang

Recent advances in generative models have demonstrated remarkable capabilities in producing high-quality images, but their reliance on large-scale unlabeled data has raised significant safety and copyright concerns. Efforts to address these…

计算机视觉与模式识别 · 计算机科学 2025-07-21 Yang Zhang , Er Jin , Yanfei Dong , Yixuan Wu , Philip Torr , Ashkan Khakzar , Johannes Stegmaier , Kenji Kawaguchi

Removing unwanted concepts from large-scale text-to-image (T2I) diffusion models while maintaining their overall generative quality remains an open challenge. This difficulty is especially pronounced in emerging paradigms, such as Stable…

计算机视觉与模式识别 · 计算机科学 2025-01-03 Daiheng Gao , Shilin Lu , Shaw Walters , Wenbo Zhou , Jiaming Chu , Jie Zhang , Bang Zhang , Mengxi Jia , Jian Zhao , Zhaoxin Fan , Weiming Zhang

Large-scale flow matching models have achieved strong performance across generative tasks such as text-to-image, video, 3D, and speech synthesis. However, aligning their outputs with human preferences and task-specific objectives remains…

机器学习 · 计算机科学 2026-03-10 Zexiang Liu , Xianglong He , Yangguang Li

Text-to-image diffusion models have demonstrated remarkable capabilities in generating high-quality images, yet their tendency to reproduce undesirable concepts, such as NSFW content, copyrighted styles, or specific objects, poses growing…

计算机视觉与模式识别 · 计算机科学 2026-02-03 Zhiqi Zhang , Xinhao Zhong , Yi Sun , Shuoyang Sun , Bin Chen , Shu-Tao Xia , Xuan Wang

Diffusion models and flow matching have demonstrated remarkable success in text-to-image generation. While many existing alignment methods primarily focus on fine-tuning pre-trained generative models to maximize a given reward function,…

机器学习 · 统计学 2026-02-03 Yidong Ouyang , Liyan Xie , Hongyuan Zha , Guang Cheng

Video generation models trained on heterogeneous data with likelihood-surrogate objectives can produce visually plausible rollouts that violate physical constraints in embodied manipulation. Although reinforcement-learning post-training…

计算机视觉与模式识别 · 计算机科学 2026-05-15 Zhenyang Ni , Yijiang Li , Ruochen Jiao , Simon Sinong Zhan , Sipeng Chen , Zhenfei Yin , Minshuo Chen , Philip Torr , Zhaoran Wang , Qi Zhu

Removing undesired concepts from large-scale text-to-image (T2I) and text-to-video (T2V) diffusion models while preserving overall generative quality remains a major challenge, particularly as modern models such as Stable Diffusion v3,…

计算机视觉与模式识别 · 计算机科学 2026-03-03 Zhaoxin Fan , Nanxiang Jiang , Daiheng Gao , Shiji Zhou , Wenjun Wu

Group Relative Policy Optimization (GRPO) has proven highly effective in enhancing the alignment capabilities of Large Language Models (LLMs). However, current adaptations of GRPO for the flow matching-based image generation neglect a…

机器学习 · 计算机科学 2025-12-16 Yawen Shao , Jie Xiao , Kai Zhu , Yu Liu , Wei Zhai , Yang Cao , Zheng-Jun Zha

Unified models capable of interleaved generation have emerged as a promising paradigm, with the community increasingly converging on autoregressive modeling for text and flow matching for image generation. To advance this direction, we…

计算机视觉与模式识别 · 计算机科学 2026-03-25 Jie Liu , Zilyu Ye , Linxiao Yuan , Shenhan Zhu , Yu Gao , Jie Wu , Kunchang Li , Xionghui Wang , Xiaonan Nie , Weilin Huang , Wanli Ouyang

Recent advances in text-to-image diffusion models have demonstrated remarkable generation capabilities, yet they raise significant concerns regarding safety, copyright, and ethical implications. Existing concept erasure methods address…

计算机视觉与模式识别 · 计算机科学 2026-04-02 Yi Sun , Xinhao Zhong , Hongyan Li , Yimin Zhou , Junhao Li , Bin Chen , Xuan Wang

Reinforcement learning (RL), particularly GRPO, improves image generation quality significantly by comparing the relative performance of images generated within the same group. However, in the later stages of training, the model tends to…

计算机视觉与模式识别 · 计算机科学 2025-12-29 Henglin Liu , Huijuan Huang , Jing Wang , Chang Liu , Xiu Li , Xiangyang Ji
‹ 上一页 1 2 3 10 下一页 ›