wuuu

心跳信号分类本文在划分数据集时，保证八万条训练样本与两万条测试样本中各个类别 的样本比例相同。处理样本类别不平衡时，仅对训练样本做不平衡处理，测试 样本不做处理，这能更全面地评价训练集中学习的分类模型的泛化性能Heartbeat Signal Classification: In this work, when splitting the dataset, we ensure that the class distribution of the 80,000 training samples and 20,000 test samples is identical across all categories. When addressing class imbalance, we only perform imbalance mitigation operations on the training samples without processing the test samples. This allows for a more comprehensive evaluation of the generalization performance of the classification model trained on the training dataset.

搜集汇总

背景与挑战

背景概述

该数据集名为'wuuu'，专注于心跳信号分类任务，包含8万条训练样本和2万条测试样本，并保持类别比例一致。仅对训练样本进行不平衡处理，以评估分类模型的泛化性能，数据文件train.xlsx大小为89.96MB。