AI for every developer - Update Conference Prague 2019 course material

Course material for the AI for every developer course. This is the repository with all the assets out of my speech at the Update Conference 2019 in Prague. You can find the slides on the assets folder both in PDF and Power Point format

• Requirements
• Cloning the repository
• Installing and configuring Anaconda -- Download and Install Anaconda Python 3.7 -- Open an Anaconda prompt and change to course folder -- Create the course environment -- Activate the environment -- Launch Jupyter Notebook -- Run the Check environment Notebook -- Troubleshooting installation -- Updating Conda
• Samples -- MNIST_MLdotNet -- MNIST_TensorFlowdotNet -- MNIST_KerasdotNet -- Notebooks -- MNIST_Client
• Useful Resources
• Credits

Requirements

To run all the samples you're gonna need Visual Studio 2017 or 2019 (any version) to run all the .Net related samples. You also gonna need .Net Core 3.0 installed on your machine. To run the ONNX converter you need Docker installed on your system.

To run the notebooks and the Python/TensorFlow/Keras related samples you're gonna need Anaconda

Cloning the repository

Open a Bash prompt, move on the parent directory where you want to clone the repository and type the following command on a Bash console:

git clone https://github.com/gsantopaolo/UpdateConference2019.git

Installing and configuring Anaconda

Download and Install Anaconda Python 3.7

https://www.anaconda.com/download/

Open an Anaconda prompt and change to course folder

cd UpdateConference2019

Create the course environment

if you want to use the CPU

conda env create

If you have an NVidia GPU GTX/RTX

conda env create -f environment-gpu.yml

wait for the environment to create.

Activate the environment

conda activate ai_for_every_dev

Check that your prompt changed to

(ai_for_every_dev) $

Launch Jupyter Notebook

jupyter notebook

Run the Check environment Notebook

Go to the notebooks folder, open the notebook check_environment.ipynb and run it. If you see the message:

Houston we are go!

You are good to go! Enjoy!

Troubleshooting installation

If for some reason you don't see Houston we are go!, the simplest solution is to delete the environment and start from scratch again.

To remove the environment:

• close the browser and go back to your terminal
• stop jupyter notebook (CTRL-C)
• deactivate the environment:

conda deactivate

• delete the environment:

conda env remove -n ai_for_every_dev

• restart from environment creation and make sure that each steps completes till the end.

Updating Conda

One thing you can also try is to update your conda executable. This may help if you already had Anaconda installed on your system.

conda update conda

Samples

MNIST_MLdotNet

Contains the sample demonstrating how to train a model to classify hand written digits using the MNIST data-set and ML.NET

MNIST_TensorFlowdotNet

Contains the sample demonstrating how to train a model to classify hand written digits using the MNIST data-set and SciSharp/TensorFlow.NET

MNIST_KerasdotNet

Contains the sample demonstrating how to train a model to classify hand written digits using the MNIST data-set and SciSharp/Keras.NET

Notebooks

Here you can find a TensorFlow notebook and a Keras Notebook demonstrating how to train a model to classify hand written digits using the MNIST data-set. You need to install the Anaconda environment and then launch corresponding the Jupyter notebook

MNIST_Client

Useful Resources

• [Book] Make Your Own Neural Network - Rashid, Tariq
• [Book] Deep Learning with Python - Chollet, Francois
• [Book] Andrew Glassner - Deep Learning, Vol. 1: From Basics to Practice
• [Book] Yaser S. Abu-Mostafa – Learning from data (inside the book instruction to access whole semester course gave by prof. Yaser S. Abu-Mostafa at California Institute of Technology)
• [Video] Andrew Glassner – Deep Learning: A crash Course
• MIT - Massachusetts Institute of Technology – Professional Certificate Program in Machine Learning & Artificial Intelligence
• Microsoft’s official AI site for devs
• Microsoft’s AI School
• [Video] Jason Mayers – Machine Learning in 45 Minutes
• [Video] Chris Lauren – Seth Juarez - Demystifying Machine and Deep Learning for - Developers
• Introducing ML.NET
• ONNX
• ONNX Model Zoo
• ONNX models for WinML
• Azure AI Gallery curates models for use with Windows ML
• Ml .Net Samples
• [Video series] CH9 – AI Show
• The MNIST Database
• Kaggle Datasets database
• Getting started with TensorFlow
• TensorFlow model zoo
• Cafee model zoo
• Christopher Olah’s blog about DL http://colah.github.io/

Credits

Special thanks to the following people and groups that contributed inspiration, ideas or support for this ever-evolving talk / workshop:

• Jason Mayes
• Rick Barraza
• Seth Juarez
• Ankit
• Chris Olah
• The Asimov Institute
• Andrew Glassner
• The Emerging Experiences friends -- Andras Velvart -- Andreas Erben -- Bronwen Zande -- David Renton -- Dennis Vroegop -- Dwight Goins -- Gordon McTaco -- James Ashley -- Jasper Brekelmans -- Joshua Blake -- Tim Huckaby -- Vincent Guigui