Jasonnor/Backpropagation

Provides a Java Swing GUI for training and visualizing multilayer neural networks, automatically splitting input datasets into 66/34 training/testing partitions. The implementation learns hyperplane decision boundaries and synaptic weights through standard backpropagation, with real-time graphical rendering of learned classification regions. Supports custom multi-dimensional datasets in simple tab-separated format and includes adjustable network parameters for experimentation.

251 stars. No commits in the last 6 months.