codenamewei / anomaly-detection-with-use-case

Modelling of unbalanced data in various use cases

Anomaly Detection using DL4J Framework

Data modelling for anomaly detection in various use cases.
The goal is to achieve faster inspection to narrow down scenarios different from the usual, for further inspection and monitoring.

Use Cases

Refer to the readme in every sub-directory for a more through elaboration.

1. Credit Card Transaction Fraud Detection:

   This use case shows the modelling of Credit Card Transaction Data using LSTM Autoencoder.
   The dataset used is large with near to 300 thousand datapoints, with normal data to fraud data having a ratio of 577:1.
   The results after 1 epoch is promising. Over 80% of frauds is identified correctly.

2. Paper Sheet Break Time Of Occurence Prediction

   This use case demonstrate the identifying of sheet breaks in paper manufacturing industry.