/Input-Invex-Neural-Network

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

Input-Invex-Neural-Network

Paper

Supplementary Material

The experiments and plots were conducted with the code (*.ipynb files).

Experiment: K-Lipschitz

The experiment comparing the capacity of various K-Lipschitz constraints method was conducted by file: 00.2_2D_GP_vs_LP_vs_SN_vs_GCLP-Benchmark-All.ipynb

Experiment: Invex

The experiment comparing the capacity of Invex function alongside Linear, Convex and Ordinary Neural Networks was conducted by:

2D Regression: 01.4_2D_reg_Linear_vs_Convex_vs_Invex_vs_NN_Benchmark.ipynb

2D Classification: 01.3_2D_cls_Logistic_vs_Convex_vs_Invex_vs_NN_Benchmark.ipynb

MNIST : MLP 05.1_Logistic_vs_Convex_vs_Invex_vs_NN_Mnist-Benchmark.ipynb

MNIST : CNN 05.2_Logistic_vs_Convex_vs_Invex_vs_NN_MnistCNN-Benchmark.ipynb

F-MNIST : CNN 05.3_Convex_vs_Invex_vs_NN_FMnistCNN-Benchmark.ipynb