cwfparsonson / AMLS_II

This code is for use within the UCL Electronic Engineering AMLS II module (ELEC0135).

AMLS II

This code is for use within the UCL Electronic Engineering AMLS II module (ELEC0135).

Setup

If using Google Colab to run the code, do the following steps:

Open the .ipynb file that you would like to run in GitHub. Then, in the URL above, change github.com to githubtocolab.com and press Enter.

In the first cell, clone this GitHub project, and run:

!git clone https://github.com/cwfparsonson/AMLS_II

Go to the directory where the file you initially opened is saved, using %cd.

For example::

If you opened the file Lab1_MLP.ipynb initially. You add a code line at the top of the code to clone the repository. Then, you add a new code cell that you use to enter the corresponding directory, %cd AMLS_II/Lab1/Lab1_MLP/

Now, in a new cell at the top of the code, run the following lines to install the needed packages:

For Lab2:
    !pip install --upgrade tf_slim

If using Anaconda virtual environment manager to run the code, do the following steps:

Download Git using the following link:

https://git-scm.com/downloads

Change the current working directory to the location where you want to clone this GitHub project, and run:

git clone https://github.com/cwfparsonson/AMLS_II

If you have installed Anaconda, you can create a new environment with Python 3.6 called amls2 by running:

conda create --name amls2 python=3.6

In your Python 3.6 environment or machine, from the route directory of where you cloned this project, install the required packages by running:

For Lab1 - MLP:
    pip install tensorflow==1.12.0
    pip install keras==2.2.4
    pip install pillow
    pip install opencv-python
    conda install -c conda-forge dlib

For Lab1 - MoE:
    pip install tensorflow==1.12.0
    pip install keras==2.2.4
    pip install seaborn==0.7.1
    pip install pydot==1.4.1

For Lab2:
    pip install tensorflow==1.12.0
    pip install matplotlib
    pip install dask --upgrade

To test that your packages have installed correctly, open Lab1/Lab1_MLP/lab1_MLP.ipynb in a Jupyter Notebook and try to run the cells.

If everything has been correctly installed, you should be able to run all Jupyter Notebook scripts in each of the folders.

Note that if you are using windows, please uncomment line 112 in Lab1/Lab1_MLP/lab3_data.py

If using Jupyter Online to run the code, do the following steps:

In the first cell, clone this GitHub project, and run:

!git clone https://github.com/cwfparsonson/AMLS_II

Open the binder folder and open the .ipynb file you want to run.

Now, in a new cell at the top of the code, run the following lines to install the needed packages:

For Lab1 - MLP:
    !pip install tensorflow==1.12.0
    !pip install keras==2.2.4
    !pip install pillow
    !pip install opencv-python
    !pip install opencv-python-headless
    !conda install -y -c conda-forge dlib

For Lab1 - MoE:
    !pip install tensorflow==1.12.0
    !pip install keras==2.2.4
    !pip install seaborn
    !pip install pydot==1.4.1
    !pip install 'h5py==2.10.0' --force-reinstall

For Lab2:
    !pip install tensorflow==1.12.0
    !pip install matplotlib
    !pip install dask --upgrade

Note that if you are using windows, please uncomment line 112 in Lab1/Lab1_MLP/lab3_data.py

Issues

The following issues have previously been encountered and resolved:

• Problems with dlib: For dlib==19.16.0 to install, you may need to separately install cmake by running python -m pip install cmake. For cmake to work, you may also need to install a C++ compiler with sudo apt-get install g++ (Linux). You should then be able to run python -m pip install dlib==19.16.0
• Jupyter accessing environment: For your environment to be selectable as a kernel in Jupyter Notebook, once you have installed the required packages into your virtual environment called <env_name>, you may need to run python -m ipkykernel install --user --name <env_name> --display-name "<env_name>" so that you can select your <env_name> in the Jupyter Notebook under Kernel -> Change kernel -> <env_name>.
• Linux memory errors: If you are partitioning your drive to run Linux, you may encounter tmp memory errors when installing the requirements.txt file. To solve this, you will need to free up swap memory on your Linux machine so the packages can be installed: https://askubuntu.com/questions/178712/how-to-increase-swap-space

If you have any questions or issues, please do not hesitate to raise a GitHub issue on this AMLS_II repository. To raise an issue, at the top of the window at https://github.com/cwfparsonson/AMLS_II, go to Issues -> New issue, and fill out your post. We will answer your issues/questions ASAP.