/detectionsc

Modelling System Call Traces for Intrusion Detection

Primary LanguagePythonMIT LicenseMIT

detectionsc (Intrusion Detection for System Calls)

Python Scripts for Formatting Data -- Using the ADFA-LD dataset which contains system call traces.

  • Scripts format the data for evluation using scikitlearn
  • Contains a model builder for a support vector machine using the ngrams

#Updates

  • Created a primitive python script to format the normal training data into the arff file
  • Began using scikit_learn's libraries to implement the corrected way of using ngramas in classification
  • Successfully implemented SVM to be able to classify ngrams
  • Moved onto only using SVMs as a means of classification due to performance
  • Implemented cross validation with the SVM implementation
  • Created a formater for compression versus uncompressed system call traces using ngrams

#TO-DO

  • Implement on a live system to validate results from the model
  • Optimization of the classification process during model building
  • See the results of having skips in the ngrams