Layer-Classifier

App for extraction of knowledge from the data set to determine the type of layer (collector, tire) using Back Propagation and Resilient Propagation Neural Nets Learning Algorithms.

Build and Run

To build this application you need gradle build tool Download Gradle To run this application you need Java SE 1.8.0_40+ Download Java

Navigate to project parent folder and run:

gradle build

To build the project. And

 java -jar build/libs/Layer-Classifier[version].jar

To run it.

Main Window

Main window contains two tables with real and training data, selectbox with neural net types and text area for results logging. To start training you should first fill training data table.

Training Settings

When you click on train button - training settings popup menu shows up. In this menu you can customize learning speed, max iterations and max available error of neural net.

Error decreasing Graphs

After training, error decreasing chart shows up. This chart represent value of error on each iteration.

Artificial Value Generation

You can write real data by hand or generate artificial values to test neural net. To generate artificial values just click on generate button and then write amount of rows which should be generated.

Classifying

When you press solve button neural net classifies layers on tire or collector (Don't forget to train neural net before run classification).

License

Layer-Classifier is released under the MIT License.