MONSTER
收藏资源简介:
MONSTER是由莫纳什大学推出的一个大规模时间序列分类数据集,旨在补充现有的UCR和UEA时间序列分类仓库。该数据集包含了29个单变量和多变量数据集,涵盖了音频、卫星图像、脑电图、人类活动识别、计数和其他类别,共有10,299至59,268,823个时间序列不等。数据集以.npy和.csv格式提供,并带有5折交叉验证的索引,以便于研究者直接进行比较。这些数据集大多已有公开可用,本文对其进行了处理后统一格式,以降低研究门槛。该数据集的推出有望推动时间序列分类领域的研究发展。
MONSTER is a large-scale time series classification dataset developed by Monash University, intended to supplement the existing UCR and UEA time series classification repositories. This dataset includes 29 univariate and multivariate datasets covering audio, satellite imagery, electroencephalography (EEG), human activity recognition, counting and other categories, with the number of time series ranging from 10,299 to 59,268,823. The datasets are provided in .npy and .csv formats, along with pre-defined 5-fold cross-validation indices to allow researchers to conduct direct comparative evaluations. Most of these datasets were previously publicly available, and this work has standardized their formats to lower the research threshold. The release of this dataset is expected to advance research progress in the field of time series classification.




