Interactive VisuApps for AI and Machine Learning

The Idea

Machine learning algorithms are often abstract and hard to grasp – especially in teaching. The VisuApps make exactly that visible: they visualise AI and ML methods interactively in the browser, without installation, without prior knowledge. Using the app means understanding the algorithm.

I am the initiator and developer of the VisuApps. They emerge as a teaching and transfer project at the intersection of academia and practice – and are used in my own courses at Reutlingen University.

Available Apps

AI Models Compared

Time horizon of tasks that various AI models can solve in 50% of cases – interactive log and linear scale, based on METR data.

Artificial Neural Network (ANN)

A simple neural network recognises handwritten digits in real time. Weights, activations and learning progress are visualised live.

K-Nearest Neighbors (kNN)

Interactive 2D visualisation of the kNN algorithm. Data points can be moved, added and deleted – classification regions update in real time.

K-Nearest Neighbors 3D

The kNN algorithm in three dimensions – rotatable, zoomable, intuitively explorable. Includes a live gesture recognition demo (rock, paper, scissors).

Support Vector Machines (SVM)

Visualisation of one of the most powerful supervised learning methods. Shows how support vectors determine the optimal decision boundary between classes.

Decision Tree

A decision tree trained on data and displayed graphically. Every decision path can be traced step by step.

Naive Bayes

Classification with the Naive Bayes algorithm – well known from spam filters. Shows how categorical features are used probabilistically to assign class labels.

Linear Regression

The simplest ML method made interactive. Data points can be placed and moved freely – the regression line adjusts immediately.

Visit the Overview

steinbeis-analytics.de/visuapps/ →