Neural Network Learning Platform

Welcome!

This interactive web platform was developed as part of a bachelor’s thesis to support the understanding of neural networks.

What's coming

You will be guided through four interactive examples.

Each example introduces additional concepts and allows you to interact with a neural network. Follow the instructions displayed on each page and experiment with the available controls.

Quick Overview

This is a visual representation of a very simple neural network.

Neural networks are built with neurons arranged in layers. Every neural network has at least an input and an output layer.
The neural network receives information from the input layer over its connections to the output layer and produces an output.

This neural network consists of an input layer with 2 neurons and an output layer with 1 neuron.

Quick Overview

By clicking on an neuron that is not in the input layer you open the edit menu for the neuron.

By dragging the sliders you can change the bias of the neuron (that is its willingness to activate) and the weights of the incoming connections (they determine how strongly the connection influences the neuron).

Quick Overview

This chart shows several colored dots. The position of each dot is described by its horizontal and vertical coordinate.

The neural network tries to decide which color belongs at every position in the chart. The background shows its current decision.

The line or area where the background changes from one color to another is called the decision boundary.

In the following examples, your task is to adjust the weights and the biases.

Try to make the background color match the color of the dots as closely as possible. By doing this, you are training the neural network by hand.