testDesignShiny: Default hyperparameter experimental test design for a...

In SpencerButt/IDPS-LAAD: Intrusion Detection Prevention System Log Autoencoder Anomaly Detector

Description

Default hyperparameter experimental test design for a single-hidden layer undercomplete autoencoder neural netowrk.

Usage

testDesignShiny

Format

A dataset with 600 rows and 12 features; 9-test design features, 1-response variable feature, and 1-design point index feature

• Subset_Split: Categorical 3-level factor; describes the subset to use in the Autoencoder

• Design_Point: An index of each test point

• Activatino_Function: Categroical 2-level factor; describes the activation function within each neuron in the neural network

• Input_DO_Rate: Continuous multi-level factor; describes the Dropout rate of the input layer neurons

• Hidden_DO_Rate: Continuous multi-level factor; describes the Dropout rate of the hidden layer neurons

• Initial_Weight_Distribution: Categorical 3-level factor; describes the distribution by which the initial weights of the autoencoder neural network are generated

• Data_Scale: Categorical 3-level factor; describes the range to which the training and test subset data is scaled.

• Rho: Continuous multi-level factor; describes the value of the rho parameter for the ADADELTA learning procedure

• Epsilon: Continuous multi-level factor; describes the value of the epsilon parameter for the ADADELTA learning procedure

• Shuffle_Train_Data: Categorical 2-level factor; boolean value indicating if the training data should be randomly shuffled during neural network training

• Y: Placeholder for the response value measurement

Information

A .rda dataset containing the factors and levels of 600-test trial designed experiment generated in JMP Pro v12.1. The designed experiment is a flexible space filling design for 9 test design factors.

SpencerButt/IDPS-LAAD documentation built on April 20, 2020, 8:45 p.m.