This is the team project on which I had worked on at RIT(Rochester Institute of Technology).

Team Members: Saunil Desai, Ravina Dandekar, Pratik Mehta

We have Designed and developed an anomaly and misuse based intrusion detection system using neural networks.

Technologies used: Java Weka and R

Java is used to prepare datasets. R is used to implement a neural network. Weka is used for data cleaning.

We created two files. One to detect anomaly based attacks and other to detect misuse based attacks.

These files have around 4500 instances. The input is divided into a Training Data Set (75%) and Test Data Set (25%).

Data Preprocessing

1. It is carried out using Weka. For details on how the data was preprocessed refer Page 4 of the report.pdf file.

Data Preparation

1. DataPreparation.java file is used to prepare data. Data is prepared by extracting rows randomly from each of the files named as Optimized_'name_of_attack' in the data folder. For eg: Optimized_FTPWrite.
2. To run the code in DataPreparation.java file. Open it and set the String value for FILE_LOCATION attribute on line 11 to the directory where the attack files would be stored as .../data/ .
3. The Excel files generated after data preparation are Dataset_Anomaly and Dataset_Misuse which would later be used by the neural network.

Neural Network

1. Change the "LOCATION_TO_DATA_FOLDER" on line 3 in the anomalySVM.R and misuseSVM.R to the location of the project folder to use it.
2. Then you are ready to use R to run the neural network and generate results.

For details on results and any specific information please refer the report.pdf file.