Neural Network Libraries - Examples

This repository contains working examples of Neural Network Libraries. Before running any of the examples in this repository, you must install the Python package for Neural Network Libraries. The Python install guide can be found here.

Before running an example, also run the following command inside the example directory, to install additional dependencies:

cd example_directory
pip install -r requirements.txt

Docker workflow

• Our Docker workflow offers an easy installation and setup of running environments of our examples.
• See this page.

nnabla-examples-utils (neu)

neu can now be installed as a python package. It provides a wide range of utility functions. For installation and usage, check utils

Interactive Demos

We have prepared interactive demos, where you can play around without having to worry about the codes and the internal mechanism. You can run it directly on Colab from the links in the table below.

Vision: Generation, Enhancement, Animation

Name	Notebook	Task	Example
SLE-GAN		Image Generation
First Order Motion Model		Facial Motion Transfer
Zooming Slow-Mo		Video Super-Resolution
StyleGAN2		Image Generation
Deep-Exemplar-based-Video-Colorization		Video Colorization
TecoGAN		Video Super-Resolution
ESR-GAN		Super-Resolution
Self-Attention GAN		Image Generation
StarGAN		Image Translation
DCGAN		Image Generation

Vision & Language

Name	Notebook	Task	Example
CLIP		Zero-shot image classification

Vision: Recognition

Name	Notebook	Task	Example
CenterNet		Object Detection
PSMNet		Stereo Depth Estimation
Face Alignment Network		Facial Keypoint Detection
YOLO v2		Object Detection
ResNet/ResNeXt/SENet		Image Classification

Audio

Name	Notebook	Task	Example
D3Net		Music Source Separation
X-UMX		Music Source Separation

Machine Learning

Name	Notebook	Task	Example
Out-of-Core training		Out-of-Core training
MixUp / CutMix / VH-Mixup		Data Augmentation
Virtual Adversarial Training		Semi-Supervised Learning
SiameseNet		Feature Embedding
Variational Auto-encoder		Unsupervised Learning

eXplainable AI

Name	Notebook	Task	Example
Grad-CAM		Visualization
SHAP		Visualization
Attention Branch Network		Visualization

Fairness of Machine Learning

Name	Notebook	Task	Example
Demographic parity Disparate Impact Equal opportunity Equalised odds		[Metrics tutorial] Dataset/Model Bias Check
Reweighing		[Pre-processing tutorial] Dataset/Model Bias Check and Mitigation by Reweighing
Massage Data		[Pre-processing tutorial] Dataset/Model Bias Check and Mitigation by Massage Data
Preferential Sampling		[Pre-processing tutorial] Dataset/Model Bias Check and Mitigation by Preferential Sampling
GAN Data Debiasing		[Pre-processing tutorial] Dataset/Model Bias Check and Mitigation by GAN
Prejudice Remover Regularizer		[In-processing tutorial] Model Bias Check and Mitigation by Prejudice Removal Technique
Prejudice Remover Regularizer for Images		[In-processing tutorial] Model Bias Check and Mitigation by Prejudice Removal Technique for Images
Adversarial Debiasing Tutorial		[In-processing tutorial] Model Bias Check and Mitigation by Adversarial Debiasing
Adversarial Debiasing for Images		[In-processing tutorial] Model Bias Check and Mitigation by Adversarial Debiasing for Images
Rejection Option based Classification		[Post-processing tutorial] Prediction Bias Check and Mitigation by ROC
Rejection Option based Classification for Images		[Post-processing tutorial] Prediction Bias Check and Mitigation by ROC for Images
Skin color (Masked Images)		Facial evaluation for skin color

Model Quantization

Name	Notebook	Task	Example
Post-training quantization		Post-training quantization