arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.03208cs.LGcs.CV
收藏

谱演化搜索:用于奖励对齐图像生成的高效推理时缩放

Spectral Evolution Search: Efficient Inference-Time Scaling for Reward-Aligned Image Generation

School of Data Science and Engineering, East China Normal University, Shanghai, China(数据科学与工程学院,东华大学,上海,中国);Alibaba Group, Hangzhou, China(阿里巴巴集团,杭州,中国)

Jinyan Ye, Zhongjie Duan, Zhiwen Li, Cen Chen, Daoyuan Chen, Yaliang Li, Yingda Chen

AI总结:

本文提出谱演化搜索(SES),通过在低频子空间内执行梯度自由演化搜索,提升奖励对齐图像生成的效率和质量。

摘要:

推理时缩放提供了一种灵活的范式,用于将视觉生成模型与下游目标对齐,而无需参数更新。然而,现有方法在优化高维初始噪声时存在严重的效率问题,因为许多搜索方向对最终生成几乎没有影响。我们表明,这种低效性与生成动力学中的频谱偏见密切相关:模型对初始扰动的敏感性随着频率的增加而迅速降低。基于这一见解,我们提出了谱演化搜索(SES),一种用于初始噪声优化的即插即用框架,它在低频子空间内执行梯度自由的演化搜索。理论上,我们推导出谱缩放预测,从扰动传播动力学中解释了扰动在不同频率下的系统性差异。广泛的实验表明,SES显著推进了生成质量与计算成本的帕累托前沿,在等效预算下一致优于强大的基线方法。

英文摘要:

Inference-time scaling offers a versatile paradigm for aligning visual generative models with downstream objectives without parameter updates. However, existing approaches that optimize the high-dimensional initial noise suffer from severe inefficiency, as many search directions exert negligible influence on the final generation. We show that this inefficiency is closely related to a spectral bias in generative dynamics: model sensitivity to initial perturbations diminishes rapidly as frequency increases. Building on this insight, we propose Spectral Evolution Search (SES), a plug-and-play framework for initial noise optimization that executes gradient-free evolutionary search within a low-frequency subspace. Theoretically, we derive the Spectral Scaling Prediction from perturbation propagation dynamics, which explains the systematic differences in the impact of perturbations across frequencies. Extensive experiments demonstrate that SES significantly advances the Pareto frontier of generation quality versus computational cost, consistently outperforming strong baselines under equivalent budgets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑