Machine Number Sense (MNS)

Machine Number Sense (MNS) 数据集是由加州大学洛杉矶分校视觉、认知、学习和自主中心创建的，旨在评估机器对抽象数字符号的理解和基于上下文的解决问题能力。该数据集包含多种类型的视觉算术问题，其中整数作为问题内容，几何形状作为问题上下文。数据集的创建过程涉及使用And-Or图（AOG）自动生成视觉算术问题。MNS数据集的应用领域主要集中在机器智能的数学思维和智能评估，旨在解决机器在理解和处理数字概念及关系操作方面的挑战。The Machine Number Sense (MNS) dataset was developed by the Center for Vision, Cognition, Learning and Autonomy at the University of California, Los Angeles, with the goal of evaluating machines' understanding of abstract numerical symbols and their context-based problem-solving abilities. This dataset encompasses a wide range of visual arithmetic problems, where integers serve as the problem content and geometric shapes act as the problem context. The construction of the MNS dataset involves automatically generating visual arithmetic problems using the And-Or Graph (AOG). The main application domains of the MNS dataset focus on mathematical thinking and intelligence assessment for machine intelligence, aiming to address the challenges that machines encounter when understanding and processing numerical concepts and relational operations.