Anomaly Detection of Ecological Monitoring Data Based on Long and Short-Term Memory Network-Autoencoder

Résumé fourni par la source

Time series anomaly detection involves identifying data points or subsequences in time series data that deviate from normal patterns. This task has significant applications in various fields, including industry, healthcare, and finance. However, detecting anomalies in time series data can be challenging due to their rarity and unknown nature, making it difficult to obtain effective labels for supervised learning. To tackle this issue, the paper proposes an unsupervised method for detecting anomalies in time series data using LSTM (Long Short-Term Memory) and Autoencoders. The method calculates the degree of anomaly by measuring the reconstruction error between input and output data. It only requires the training of normal data, without the need for labeling anomalies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.