arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2209.15087cs.CVcs.AI
收藏

基于概率类比映射的Zero-shot视觉推理

Zero-shot visual reasoning through probabilistic analogical mapping

University of California, Los Angeles(加州大学洛杉矶分校);Air Force Research Laboratory(空军研究实验室)

Taylor W. Webb, Shuhao Fu, Trevor Bihl, Keith J. Holyoak, Hongjing Lu

摘要:

人类的推理是建立在识别高度抽象的共性的能力之上的,这些共性支配着表面上不同的视觉输入。最近开发具有这种能力的算法的努力主要集中在需要对视觉推理任务进行广泛的直接训练的方法上,并且对具有新内容的问题的泛化能力有限。相比之下,认知科学的长期研究传统一直专注于阐明人类类比推理的计算原理;然而,这项工作通常依赖于手动构建的表示。在这里,我们提出了视觉概率类比映射(visual Probabilistic Analogical Mapping,简称PAM),一个综合了这两种方法的视觉推理模型。VisiPAM采用直接从自然视觉输入中获得的学习表示,再加上从人类推理的认知理论中获得的基于相似性的映射操作。我们表明,在没有任何直接训练的情况下,CNOPAM在类比映射任务上的表现优于最先进的深度学习模型。此外,CNOPAM与人类在一项涉及跨不同类别映射3D对象的新任务上的表现模式密切匹配。

英文摘要:

Human reasoning is grounded in an ability to identify highly abstract commonalities governing superficially dissimilar visual inputs. Recent efforts to develop algorithms with this capacity have largely focused on approaches that require extensive direct training on visual reasoning tasks, and yield limited generalization to problems with novel content. In contrast, a long tradition of research in cognitive science has focused on elucidating the computational principles underlying human analogical reasoning; however, this work has generally relied on manually constructed representations. Here we present visiPAM (visual Probabilistic Analogical Mapping), a model of visual reasoning that synthesizes these two approaches. VisiPAM employs learned representations derived directly from naturalistic visual inputs, coupled with a similarity-based mapping operation derived from cognitive theories of human reasoning. We show that without any direct training, visiPAM outperforms a state-of-the-art deep learning model on an analogical mapping task. In addition, visiPAM closely matches the pattern of human performance on a novel task involving mapping of 3D objects across disparate categories.

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。

↑