Synthetic CBAI Instances
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/lasgroup/adaptive-constraint-learning
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是为了评估ACOL算法并在受限线性最优臂识别（CBAI）任务中与自然基线方法进行比较而合成的实例。数据集包含了对于学习正确约束边界无关的维度、不同的臂以及观测值中的高斯噪声。规模上，数据集具有多个维度，且臂的数量各不相同，其任务是对受限线性最优臂进行识别。
This dataset is a synthetic instance constructed to evaluate the ACOL algorithm and compare it with natural baseline methods in the Constrained Best Arm Identification (CBAI) task. The dataset includes dimensions irrelevant to learning the correct constraint boundaries, distinct arms, and Gaussian noise present in the observations. In terms of scale, the dataset features multiple dimensions and varying numbers of arms, with its target task being constrained linear best arm identification.
提供机构：
Authors of the paper
