[Event] Numpy for Machine Learning


#1

Hi! Last Saturday we had a tech session on GO. This time we are doing Numpy for Machine Learning, tomorrow 9 PM.

Numpy is the core library for scientific computing in Python. It provides a high-performance multidimensional array object, and tools for working with these arrays.

I’m planning to take a session on the basics of Machine Learning in the summer camp next month (Villupuram, Pondy). During the session, Introductory Course on Machine learning, we will be using numpy to build the models introduced in Andrew Ng’s Machine Learning course and solve the excercises. Note that octave is used in that course for the same purpose. Since most of the existing machine learning frameworks today are based on python (Tensorflow, Theano, etc), I figured it would be practical to use numpy to understand these basic machine learning models. Numpy is also easy to learn and experiment with and it plays well with Tensorflow and Theano.

So what can we do with numpy?

We certainly cannot build any powerful and intelligent models with numpy. And it is not what numpy is for. With numpy you can play around with your dataset; Twist it, turn it, analyze it and prepare it for the the gaint deep models to consume.

We will learn by practicing a few excercises in ipython. You can expect something like this. If this goes well, we’ll move on to Advanced Machine Learning for A.I. in the coming sessions.

Setup

# remove numpy 
sudo apt-get remove python-numpy
# install and update pip
sudo apt-get install python-pip
sudo -H pip install --upgrade pip
# install latest version through pip
sudo -H pip install --upgrade numpy
# install ipython notebook
sudo -H pip install --upgrade ipython ipython-notebook
# optional : matplotlib; deal with dependency issues
sudo -H pip install --upgrade matplotlib

Prerequisites

  1. Python
  2. High school level Linear Algebra knowledge (Matrices and operations)
  3. Love for A.I.

Reading material : Numpy Primer
Link for the call : https://meet.jit.si/fsftntechnights

See you tomorrow, Sunday 9 PM.


#2

Contents of this session are available in this blog post. I’ve added broadcasting and array masking which I did not cover during the session.