In this course, we learn about neural networks. Neural networks are found in our brains. We use them to think. Click on the brain to reveal what is inside.

On the left, you see an artificial neural network (ANN). It is a computer representation of a biological neural network. Our ANN has 51 perceptrons. A perceptron is a computer version of a neuron in our brains.

Before a neural network can think clearly, it has to be trained. Click on the Start Learning button at the bottom to train the network to recognize the letters A .. J. It will start very slowly. Adjust the speed with the slider at the bottom left of the window.

We show the eye a pattern, which is sent to the network through the optic nerve. The network computes an output. We then tell the ANN what the output should have been. The network then realises that it is not yet seeing what it is supposed to see and corrects its internal values by just a little bit.

The dial below will keep you posted on the learning progress. Typically, the network needs to go for about 200-300 rounds of looking at the letters A..J before it is certain enough that is gets it right.

In our ANN, information flows from left to right. The greener a line, the more a Perceptron cares about the output from a connected perceptron on the left. The redder, the more it ignores the output from its predecessor. When a line is black then it is about neutral. Just like in real life. Some people you trust, some you don't or even do the opposite of what they say.

When learning is finished, drag the slider on the left. For each letter, the neural network will show you at its output layer what it thinks that this letter is. The more black a Perceptron, the more certain the network is about its choice.