A BI-LSTM Attention Mechanism (Bi-LSTM-Attn) for Monitoring Seismic Events, Solving the issue of Interpretability

The ConvNetQuake INGV Dataset is a publicly available seismic dataset originally released with the ConvNetQuake framework developed for automatic seismic event detection and phase classification. The dataset contains continuous waveform recordings and manually labeled earthquake and noise segments collected from the Istituto Nazionale di Geofisica e Vulcanologia (INGV), Italy. It is designed for developing and benchmarking machine learning and deep learning models for seismic signal processing. The dataset is particularly valuable for exploring temporal signal modeling, feature extraction, and classification under class imbalance, as it contains a large number of low-magnitude seismic events interspersed with extensive noise segments.

ConvNetQuake INGV数据集（ConvNetQuake INGV Dataset）是一套公开可用的地震数据集，最初随专为自动地震事件检测与震相分类开发的ConvNetQuake框架（ConvNetQuake Framework）一同发布。该数据集包含从意大利国家地球物理与火山研究所（INGV）采集的连续波形记录，以及经人工标注的地震与噪声片段。其设计目的是为开发与基准测试用于地震信号处理的机器学习（machine learning）与深度学习（deep learning）模型提供支撑。该数据集尤其适用于探索类别不平衡场景下的时序信号建模、特征提取与分类任务，因其包含大量散布于海量噪声片段中的低震级地震事件。