shirleyqueen / kddcup99_nid_lstm Goto Github PK

Author: Jiachuan Deng

This is an implementation according to paper: https://www.researchgate.net/profile/Ralf_Staudemeyer/publication/279770740_Applying_long_short-term_memory_recurrent_neural_networks_to_intrusion_detection/links/559a972908ae99aa62ce1468/Applying-long-short-term-memory-recurrent-neural-networks-to-intrusion-detection.pdf
Note that Since cells in one block shares gates, we use one cell with larger hidden size to replace multiple cells in original paper.
By default model is training on 10% data, and test on 90% data. You can change the ratio in config.ini.


REQUIRED PACKAGES:
	. Tensorflow version 1.8.0
	. Sklearn version 0.19.1
	. Numpy
	. Pandas

DIRECTORY STRUCTURE:
	NID_LSTM
	| README
	|
	| data
	|	| kddcup.data 
	|
	| src
		| config.ini
		| dataprocessing.py
		| minibatcher.py
		| run_model.py
		| targetEncoder.py
		| runMe.py
		| README


FILES DESCRIPTION:
	. kddcup.data: can be downloaded from http://kdd.ics.uci.edu/databases/kddcup99
	. config.ini: you can change data path and parameters of model in this file
	. minibatcher.py: python scripts used in the main executable to load the dataset and create minibatches
	. run_model.py: define the tensorflow model graph, main executable script
	. targetEncoder.py: Implementation of Empirical Bayes method according to https://kaggle2.blob.core.windows.net/forum-message-attachments/225952/7441/high%20cardinality%20categoricals.pdf. can be used to deal with categorical features
	. runMe.py: script to run all scripts



RUNNING THE CODE:
	> run command "python3 runMe.py" in src directory.
	> all parameters can be changed in config.ini
	> Before running the code, make sure the "[DataPath] traindatapath" in config.ini is changed according to exactly where you place kddcup.data.