top of page

Reading Material

Neural Network and Deep Learning:

Dr Spyros Chatzivasileiadis suggested reading the following papers to be prepared for the course. All papers are available for free from ARXIV as listed below:

 

Reinforcement Learning:

Dr Zoltan Nagy suggested reading his recent book chapter that is accessible from this link. Please do not share this book chapter with anyone as it is protected by copyright law.

In addition, he suggested reading the Barto & Sutton book (the main reference for RL scientists and users), which is available online for free at http://incompleteideas.net/book/the-book-2nd.html

Clustering and Classification:

Dr Archie Chapman suggested reading the textbook “Pattern Recognition and Machine Learning,” by Christopher Bishop. A soft copy of the book is freely available on Prof Bishop’s webpage at https://www.microsoft.com/en-us/research/people/cmbishop/

Required Software Packages

Our speakers prepared hands-on exercises in each session. In order to be able to follow them on your laptop, you need to install the following software and packages. The installation process is explained below for Windows 10, as it is sometimes tricky to get them to work on Windows machines. However, you can install them on your MAC or Linux machines and they should work.

Python package installation guide on Windows 10

Installing Python 3.7 (Please do not install newer versions as they don’t work with TensorFlow)

  1. Download anaconda graphical installer from https://www.anaconda.com/distribution/#windows

  2. Run the installer

  3. Run Anaconda prompt

  4. Run the following commands:

    1. pip install torch==1.3.1+cpu torchvision==0.4.2+cpu -f https://download.pytorch.org/whl/torch_stable.html

    2. pip install gym matplotlib pandas numpy pathlib agent

  5. Now you’re ready to run the main.py file

PS: add the following two lines to line 4 of main.py in order to get it to run

       from agent import RL_Agents
       from reward_function import reward_function

In order to make sure that the packages are installed correctly, and as a part of requirements for “Reinforcement Learning” sessions, please follow the instructions in below:

  1. Download the repository by clicking on “Clone or download” on the Github repository

  2. Uncompress the CityLearn-Master.zip

  3. Change directory in Anaconda prompt to the uncompressed folder (cd command)

  4. Run the following command:        
    python main.py 

  5. The runtime may be different depending on the computer configuration. After code execution is completed, the following lines will be shown up.

* The file takes about 10 mins to be executed on a laptop Core i7 / 16 GB RAM / intel HD graphics 520

Tensorflow package installation guide on Windows 10

System requirements:

         Windows 7 or later (64-bit) - Python 3.4–3.7 - pip 19.0 or later 

        

Installing Anaconda:

Skip this stage if you already installed Anaconda

  1. Download anaconda graphical installer from https://www.anaconda.com/distribution/#windows

  2. Run the installer

  3. Run Anaconda prompt


Installing pip:

  1. Run the following command:

         conda install pip

You can verify that pip was installed correctly by opening a command prompt and entering the following command:

         pip --version

Installing Tensorflow:            

Install TensorFlow with Python's pip package manager by running the following command:

         conda create -yn tflow tensorflow
     conda activate tflow

 

To validate the package installation, please follow a simple example at https://www.tensorflow.org/tutorials/quickstart/beginner or simply download this Python code on your computer, change the directory on Anaconda to where the file is saved, and run it in Anaconda prompt as follows:

     python tFlow_test.py

This script trains and evaluates a neural network for MNIST dataset using Tensorflow.

The final results are given in the snapshot below.

A copy of pre-school reading materials and required software installation instruction is available in this PDF file. 

bottom of page