This Project is detect outliers in sensor networks. We are using ISSNIP Single hop dataset for this.
We have experimented with two models LSTM and SVM.
LSTM Model and training Script is present in Code/LSTM and similar for SVM. For Training we have used Mahalanobis' Distance.
Ref : https://www.renom.jp/notebooks/tutorial/time_series/lstm-anomalydetection/notebook.html
Requirements : Numpy, Scipy, sklearn, torch latest version.