andreasbinder/ML-Approach-to-reduce-the-Optimism-Bias-towards-Climate-Change

Important

The original repository can be found here. In case of in problems, please check out the repo on github. Concerning functionality, there should be no difference.

ML-Approach to reduce the Optimism Bias towards Climate Change

This project is part of a seminar at the Technical University Munich focusing on sustainable development. It's goal is to build a Generative Adversarial Network being able to generate pictures of burning houses and deploy it to the web via conversions supplied by tensorflowjs. The running prototype can be found here.

Getting started

Here's what you need to set up the project.

Prerequisites

Since this project is written in Python you will need to have a Python version >= 3.6.8 running on your system. Other you can upgrade it as follows:

sudo apt-get install python3

You can get the package manager by typing:

sudo apt-get install python3-pip
# you might need to update the package manager to activate the package
sudo apt-get update

The deep learning framework used is Keras with a Tensorflow backend. In case any problems arise, please check out the official documentation for Keras here and for Tensorflow here.

# Keras version used 2.1.1
pip3 install Keras
# Tensorflow version used 1.15.0
# Please be aware that tensorflow 2.0 might be not compatible
pip install --upgrade tensorflow==1.15

For image processing we make use of the PIL library. Please get the newest version accordingly.

pip3 install Pillow

If you wanted to deploy your own prototype to a Website, you would need to convert it to a Tensorflow.js model. The script is placed in the folder Scripts/ConvertModel.ssh. To execute the script, you need to have the tensorflowjs_converter. By installing installing tensorflowjs you will be provided with the converter.

pip3 install tensorflowjs

For more detailed instructions, please check out their homepage.

Installation

You can install the project by cloning it via https:

git clone https://github.com/andreasbinder/AppliedML_using_GAN.git

I have coded the implementation on Linux Mint 19.2 . Hence, you might need to adjust the relative paths, especially if you are a Windows user.

Test

You can run a demo of it by executing this code in the commandline when navigated into the project:

python3 prototype/Test/InstallationTest/InstallationTest.py

After few seconds you should see some generated pictures in this folder Test/InstallationTest/Results .

To experiment with the code yourself, please check out the Model directory.

Author

Andreas Josef Binder, B.Sc. Information Systems, Technical University Munich

ML-Approach-to-reduce-the-Optimism-Bias-towards-Climate-Change