Autoencoders-for-Anomaly-Detection

Google Colab Notebook for Anomaly Detection for Autoencoders

This notebook aims to test the potential of autoencoders in the realm of anomaly detection. It aims to build an autoencoder that has been trained upon the healthy and normal operation conditions of an engine, so that when it views input data that contains anomalies, it produces a high reconstruction error that can be used to flag the input as anomalous. If you'd like to view the code explanation, you can do so here:

Dataset and Research Article:

• Research Article: https://www.researchgate.net/publication/375511821_EngineFaultDB_A_Novel_Dataset_for_Automotive_Engine_Fault_Classification_and_Baseline_Results
• Dataset: https://github.com/Leo-Thomas/EngineFaultDB