cdesilv1/securitai-lstm-model

keras lstm rnn to perform binary classification on request logs.

As explained in Detecting Malicious Requests Using Keras & Tensorflow

python train.py [path to access.csv]

Training will split dataset into 75% train and 25% evaluation subsets. Model and metadata are saved upon completion.

python predict.py [request log entry]

Loads saved model created from training to output confidence on given request entry.

Requirements:

• keras @ 2.0.5
• h5py