梦想奖-X:提升高质量的3D生成与人类偏好对齐.
IEEE transactions on pattern analysis and machine intelligence
|September 15, 2025
概括
梦想奖励通过学习人类偏好来增强文本到3D生成,创建高保真性和多样化的3D模型. 这个框架,DreamReward++,通过使用一种新的奖励意识的噪音采样策略,改善了快速对齐和世代多样性.
科学领域:
- 计算机视觉 计算机视觉
- 人工智能的人工智能
- 3D 图形 3D 图形
背景情况:
- 大规模的扩散模型有先进的3D内容生成.
- 将3D内容与人类偏好保持一致,尤其是在以文本为导向的场景中,仍然是一个重大挑战.
研究的目的:
- 提出DreamReward,这是一个用于改进使用人类偏好反来改进文本驱动的3D生成的新框架.
- 开发Reward3D,一个通用的文本到3D的人类偏好奖励模型.
- 为优化3D生成模型引入Reward3D反学习 (DreamFL) 算法.
主要方法:
- 通过系统的注释管道 (过,评分,排名) 收集了超过25,000个专家的比较.
- 开发了Reward3D,这是一个人类偏好奖励模型,用于文本到3D生成.
- 实现了Reward3D反学习 (DreamFL) 算法,以将生成与用户提示保持一致.
- 将框架扩展到4D和图像到3D的生成 (DreamReward-4D,DreamReward-img).
- 拟议的DreamReward++具有奖励意识的噪音采样策略,以增强多样性和偏好调整.
主要成果:
- 在快速对齐和生成忠实性方面取得了显著的改进.
- 展示了用于4D和图像到3D生成的低成本但有效的扩展.
- DreamReward++成功地产生了高保真度,多样化的3D结果,解决了多样性的局限性.
- 奖励意识的噪音采样策略增强了文本驱动的多样性,同时保持了人类偏好调整.
结论:
- 从人类反中学习是一个强大的方法来增强3D生成模型.
- 梦想奖励及其扩展为高质量和多样化的3D内容创作提供了一个有希望的方向.
- 提出的方法显示了未来在文本驱动的3D生成方面取得重大进展的巨大潜力.
相关概念视频
Improving Translational Accuracy
3.6K
3.6K
Improving Translational Accuracy
14.1K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
14.1K
Impression Management Techniques III: Aligning Actions
125
Aligning actions are communicative strategies individuals employ to maintain social harmony and preserve personal identity in the face of potential disruptions to social norms. These actions are particularly important in managing social impressions when one's behavior might be seen as inappropriate, incompetent, or morally questionable.Types of Aligning ActionsThe three principal types of aligning actions are disclaimers, accounts, and apologies.DisclaimersDisclaimers are preventive; they are...
125
Impression Management Techniques IV: Altercasting
173
Altercasting is a strategic communication technique in which an individual imposes a specific identity or social role onto another person to influence their behavior and shape the interaction. By presuming a role—such as “responsible leader” or “patient person”—altercasting encourages the target to conform to that identity, often aligning their behavior with the expectations associated with the role. The power of this tactic lies in its subtlety; once a role...
173
Modeling and Similitude
617
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
617
Stereotype Content Model
15.3K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
15.3K


