Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 416 Bytes

README.md

File metadata and controls

6 lines (4 loc) · 416 Bytes

A Neural Networks' decision boundary visualization framework

We use Monte Carlo method to visualize the how a neural network allocates its decision space for different classes. The following figure is derived using the MNIST dataset.

Associate region of classes on MNIST dataset. Data are mapped into a 3D space using t-SNE. alt text