NNDDosForecast is a stream based hierarchical LSTM model for DDoS attack forecasting.
The input is a tweet stream, and output is the attack probability of the target in a certain day.
The codes are the implementation of following paper:
Zhongqing Wang, and Yue Zhang. DDoS Event Forecasting using Twitter Data. In Proceeding of IJCAI-17.
Following are four kinds of stream models which are described in the paper.
CNN based Vanilla Stream Model: cnn_prediction()
LSTM based Vanilla Stream Model: lstm_prediction()
Short- and Long-Term Stream Model: lstmMulti()
Hierarchical Stream Model: lstmMulti2()