sbi-dev/sbibm
收藏资源简介:
该数据集包含两个配置:bnn和bnn_reference,用于贝叶斯神经网络(BNN)或相关机器学习任务。bnn配置包含97个theta特征(可能表示模型参数)和200个x特征(可能表示输入变量),所有特征为浮点数值,总样本量110万(训练集100万,测试集10万)。bnn_reference配置包含观测ID、种子、200个观测特征和部分theta特征,可能用于参考或基准测试。数据集未提供具体应用背景,但结构暗示其适用于参数估计、模拟或概率建模。
This dataset includes two configurations: bnn and bnn_reference, likely for Bayesian Neural Networks (BNN) or related machine learning tasks. The bnn configuration features 97 theta variables (possibly model parameters) and 200 x variables (possibly input variables), all in float32 format, with a total of 1.1 million samples (1 million for training, 100,000 for testing). The bnn_reference configuration includes observation ID, seed, 200 observation variables, and partial theta variables, potentially for reference or benchmarking. No specific application context is provided, but the structure suggests use in parameter estimation, simulation, or probabilistic modeling.





