Skip to content
This repository has been archived by the owner on Mar 21, 2018. It is now read-only.
/ wksh_intro-to-nns Public archive

An Introduction to Neural Nets on MNIST, featuring Tensorflow.

License

Notifications You must be signed in to change notification settings

ucfai/wksh_intro-to-nns

Repository files navigation

Workshop: Intro to Neural Networks

v0.3.0 (Last Updated: Oct. 02, 2017)

This workshop should take about two hours to complete, with someone providing a walkthrough.

We'll cover the basics of creating a Neural Network from the ground up. We had initially planned to cover TensorFlow, but now we feel that it's beyond our own skillset to cover effectively, but also beyond the scope of what we want you to get out the workshop. If you'd like to learn more about TensorFlow talk to one of the coordinators afterwards.

  • NOTE: If you've already got docker and docker-compose on your system, then skip to the next section.
  • NOTE: Docker requires that macOS users have Yosemite or newer versions.
  • NOTE: Docker requires that Windows users have Windows 10 Pro, Education, or Enterprise.
  • NOTE: If you're ineligible, please attempt downloading Docker Toolbox.
    • If Docker Toolbox is non-functional, then, as a last resort, use conda envs (detailed below with "Using Anaconda").

Windows

Using Docker

1. Install Dependencies

  1. Install docker (Community Edition is sufficient).
  2. [Only if installing docker doesn't work]. Install Docker Toolbox.

2. Getting It Up & Running

  1. Open PowerShell.
  2. Run Set-Location -Path C:\path\to\your\cloned\repo.
  3. Run docker-compose build.
    • Do this before coming to the workshop. It will save time and headache for all participants.
  4. Run docker-compose up.
    • Use this to check that you're able to access the Jupyter environment we've setup.
  5. Open your browser and go to http://localhost:19972/.
    • If you can access this, you've got access to Jupyter. STOP if you're not at the workshop and plan on attending.
  6. Enjoy the workshop through workshop.ipynb.

Using Anaconda

1. Install Dependencies

  1. Download and install Miniconda3-4.3.21
  2. While installing Miniconda3-4.3.21, make sure you add it's location to your $PATH.

2. Getting It Up & Running

  1. Open Command Prompt.
  2. Run cd "C:\path\to\your\cloned\repo".
  3. Run conda env create -f environment.yml
    • Do this before coming to the workshop. It will save time and headache for all participants.
  4. Run activate wksh_intro-to-nns
  5. Run jupyter notebook --port=19972 --NotebookApp.token=.
    • Use this to check that you're able to access the Jupyter environment we've setup.
  6. Open your browser and go to http://localhost:19972/.
    • If you can access this, you've got access to Jupyter. STOP if you're not at the workshop and plan on attending.
  7. Enjoy the workshop through workshop.ipynb.

macOS

Using Docker

1. Install Dependencies

  1. Install docker (Community Edition is sufficient).
  2. [Only if installing docker doesn't work]. Install Docker Toolbox.

2. Getting It Up & Running

  1. Open Terminal.
  2. Run cd /path/to/your/cloned/repo.
  3. Run docker-compose build.
    • Do this before coming to the workshop. It will save time and headache for all participants.
  4. Run docker-compose up.
    • Use this to check that you're able to access the Jupyter environment we've setup.
  5. Open your browser and go to http://localhost:19972/.
    • If you can access this, you've got access to Jupyter. STOP if you're not at the workshop and plan on attending.
  6. Enjoy the workshop through workshop.ipynb.

NOTE: Using Anaconda shouldn't be necessary on macOS.

Linux

Using Docker

1. Install Dependencies

  1. Installing docker:
    • For Debian-based (Ubuntu, Mint, Debian): sudo apt update && sudo apt install -y docker
    • For Red Hat-based (Fedora, CentOS, RHEL): sudo dnf update && sudo dnf install -y docker
    • For Arch Linux: sudo pacman -Syyu && sudo pacman -S docker
  2. Installing docker-compose:
    • Check the current release version
    • curl -L https://github.com/docker/compose/releases/download/<current_version>/docker-compose-`uname -s`-`uname -m` > /usr/local/bin/docker-compose
    • chmod +x /usr/local/bin/docker-compose

2. Getting It Up & Running

  1. Open Terminal/Konsole/xterm/etc.
  2. Run cd /path/to/your/cloned/repo.
  3. Run docker-compose build.
    • Do this before coming to the workshop. It will save time and headache for all participants.
  4. Run docker-compose up.
    • Use this to check that you're able to access the Jupyter environment we've setup.
  5. Open your browser and go to http://localhost:19972/.
    • If you can access this, you've got access to Jupyter. STOP if you're not at the workshop and plan on attending.
  6. Enjoy the workshop through workshop.ipynb.

Some Extras

If you want to learn more about docker and it's intricacies, take a gander at the docs. It's an awesome tool, especially when working with multiple systems that allows you to be host-OS agnostic.