A Neural network can learn to distinguish thumbs-up and thumbs-down.

In this experiment, we train the neural network with different images showing thumbs-up and thumbs-down. Scroll through the shapes and take a look. You will first see the sequence of thumbs-up images followed by the thumbs-down images. Their order is not important. What is important is that each of the learning data images are assigned to a class, which is either Thumbs-up or Thumbs-down. You see the classes on the right-hand side of the neural network (output layer). Then, start the learning process. Hint: Press on the Hide Wires button to make the learning process faster.

Once learning is complete, drag the pattern slider. Did our ANN recognize the images correctly?

Choose one of the images and start modifying it by clicking on its pixels. A click will flip a pixel from black to white or from white to black. How much can you change an image before the neural network is no longer able to classify it correctly as either Thumbs-up or Thumbs-down?

When you hover over an artificial neuron in the output layer, you can see how certain the neural network is. Look for the dot in the graph.