Anomaly Detection Using Machine Learning

Abstract: Comparative Machine Learning Analysis on Electrocardiogram (ECG) Anomaly Detection. Anomaly Detection on the ECG dataset of 4998 patients was done with each patient having 140 data points, around 7,00,00 data points. Data is divided as 4998 patients learning data. Machine Learning (ML) model is created using Algorithms like Logistic Regression, Decision Tree Classifier, Random Forest Classifier, and Support Vector Machines on which and remaining 1000 patient data is tested to get accuracy to check whether the model has learned correctly. The same is again analyzed by Deep Learning algorithms like RMSProp, Adam Optimization, and SGD.

Keywords: Anomaly detection, Electrocardiogram (ECG), Machine Learning (ML), Deep Learning (DL), Artificial Intelligence (AI), SVM (Support Vector Machines).