Machine Learning for Software Developers (Video)
This repository contains material for a short workshop on machine learning. It is intended for software developers who have little to no experience in the field.
- Day 1 (Slides)
- What is learning?
- What is programmed vs what is learned
- A high-level look into brains (neurons and synapses)
- Code deep-dive: building a brain and teaching it to recognize Arabic digits
- Exercise 1: teaching a brain to recognize English digits
- Day 2 (Slides)
- Recap of day 1
- A gentle introduction to neural networks
- Building a neural network using Keras
- Exercise 2: building a neural network to recognize Arabic characters
The following instructions will guide you through setting up so that you have all the software and datasets needed for the workshop.
Create an account on Kaggle
You'll need this account for some of the projects we'll be doing.
- Arabic Handwritten Digits Dataset
- MNIST Dataset (download both
train.csvandtest.csv) - Arabic Handwritten Characters Dataset
2) Download and install Anaconda (Python 2.7) from here.
To set up the development environment, you'll need to do the following:
- Open
Anaconda Prompt - Run
conda create -n ml-workshop. This creates a sandbox where we'll install all the software we'll need. - Activate the environment:
activate ml-workshop - Install the needed libraries:
conda install matplotlib numpy pandas jupyter theanoconda install -c conda-forge keras
2) Download and install Anaconda (Python 2.7) from here.
To set up the development environment, you'll need to do the following:
- In your terminal run:
conda create -n ml-workshop. This creates a sandbox where we'll install all the software we'll need. - Activate the environment:
source activate ml-workshop - Install the needed libraries:
conda install matplotlib numpy pandas jupyter theanoconda install -c conda-forge keras