Connections can count differently

Look at the lines joining each input to the artificial neuron. Each line is a connection, the path along which that input affects the neuron.

So far, every active input has added 1. Every connection has counted equally.

But some inputs should matter more than others.

Give each connection a number that controls how much its input counts. That number is called a .

A high weight gives an input more influence. A low weight gives it less.

Click each input to turn it on or off. Drag the weight on its connection and watch that input's contribution change.

Here, the weights change only when you move them. Rosenblatt's Perceptron added a procedure that could change them after a mistake.

Look at the lines joining each input to the artificial neuron. Each line is a connection, the path along which that input affects the neuron.

So far, every active input has added 1. Every connection has counted equally.

But some inputs should matter more than others.

Give each connection a number that controls how much its input counts. That number is called a .

A high weight gives an input more influence. A low weight gives it less.

Click each input to turn it on or off. Drag the weight on its connection and watch that input's contribution change.

Here, the weights change only when you move them. Rosenblatt's Perceptron added a procedure that could change them after a mistake.