A Small Tool for Explaining Machine Learning
Over the last few years, I have had many opportunities to discuss artificial intelligence with a wide variety of audiences: high school students (presentation here, visitors during outreach events such as Pint of Science, colleagues from other disciplines, and of course university students.
One thing quickly became apparent: many people are curious about machine learning, but the inner workings of these algorithms often remain mysterious. Terms such as gradient descent, loss function, decision boundary, or neural network are frequently mentioned, yet they can be difficult to explain intuitively and without too much equations.
To address this and inspired by the Tensorflow Playground, I vibe coded a small interactive platform designed to make machine learning algorithms visible.
The goal is simple: instead of presenting static equations or slides, let users observe learning in action. The platform provides interactive visualisations of optimisation landscapes, classification boundaries, and neural network behaviour, allowing users to simple machine learning methods with visual intuition.
This project was initially inspired by my experiences in science outreach and teaching. I wanted a tool that could be used equally well during a public demonstration, in a classroom, or by anyone curious to better understand what happens behind the scenes of modern AI systems.
The application is open source and available online. Feel free to use it in your own courses, workshops, outreach activities, or simply for exploration. If you find it useful, do not hesitate to share it, adapt it, or contribute new ideas.
Demo and source code hosted by Hugging Face :