NNLab

Desktop application that allows to create, train and visualize neural networks. It was part of my BEng thesis.

Remaining parts of project:

• NNLib - neural network library https://github.com/pekalam/NNLib
• NNControl - WPF control displaying neural network visualization https://github.com/pekalam/NNControl

Main capabilities:

• Read data from csv, enter a set of points, normalize and visualize data
• Create and visualize neural network
• Configure training process and train neural network using high-performance library (CPU only)
• Visualize training process with 5 interactive charts updated at real-time
• View prediction results

Frameworks and libraries:

• WPF
• Prism
• Math.NET numerics (https://numerics.mathdotnet.com/)
• MahApps.Metro (https://github.com/MahApps/MahApps.Metro)

Screenshots

More screenshots: https://github.com/pekalam/nnlab/tree/master/gh_images/screenshots

Short overview of architecture

TODO

Package level overview

MVVM architecture