Chrysolites Tech Academy

Video by freeCodeCamp.org · taught by Radu Mariescu-Istodor · streamed from YouTube. All rights remain with the creator.

Lesson 3: Machine Learning Without Libraries

About this lesson

This course by Radu Mariescu-Istodor teaches machine learning with no black boxes: instead of calling a library, you write every part of the system yourself in plain JavaScript, running in an ordinary web browser with no installation. The aim is to understand exactly what happens at each stage, so that later, when you use ready-made tools, you know what they are doing for you. The project is a classifier for simple hand-drawn sketches. It starts with the data, building a small drawing app to collect labelled sketches and then working with the collected samples. You learn feature extraction, turning each drawing into a few meaningful numbers, and why those numbers need to be scaled before they can be compared fairly. A major thread of the course is visualisation: you build a viewer for the whole dataset and a custom chart from scratch, which makes it possible to see how the samples of each class are spread out and where they overlap. With the data understood, you implement a nearest-neighbour classifier and then extend it to the k-nearest neighbours classifier, seeing how each decides on a label for a new drawing. The course then turns to model evaluation, measuring how well the classifier performs on data it has not seen, and sets homework assignments to extend what you have built. Near the end it shows how the same ideas look in Python with scikit-learn. Afterwards, read Chapter 3 of the course book, which works a scaling example by hand.

Install Chrysolites Tech Academy

Android / Chrome / Edge: open the browser menu and choose Install app or Add to Home screen.

iPhone / iPad (Safari): tap the Share button, then Add to Home Screen.