/comp421-project

Feature visualization of a convolutional neural network.

Primary LanguageJupyter Notebook

Auto-encoder reconstructed images

Generative Model Regularization for Feature Visualization by Optimisation

An implementation of the ideas introduced in "Synthesizing the preferred inputs for neurons in neural networks via deep generator networks" https://arxiv.org/pdf/1605.09304.pdf

The first file you should open is notebook.ipynb It is viewable through GitHub, or Google Colab. The notebook won't run, because it depends on other python files and saved models, but it has images and details not contained in the report.