• Contribute
• What's New
• Submit News

Training a perceptron to learn the OR Logic gate is simple (and trivial!). Nevertheless, it is a great way to see how the Perceptron Learning Rule can be used.

Outline

We then compare the output of the perceptron with the desired output, and adjust the weights accordingly.

Now, we are going to train the perceptron to learn the OR-logic gate. The truth table for the OR gate is as follows:

x1	x2	y
0	0	0
1	0	1
0	1	1
1	1	1

By visualizing the graph of the truth table, it is easy to see that the line separating the points could be y=-x+1. Any line though that passes between the origin and the (0,1)(1,1)(1,0) trio will suffice though.

Guidelines

where w₁ and w₂ are the two weights, and b is the bias.

Note, that the OR logic gate was chosen simply because it is easy to see and debug. A single perceptron can classify any linearly separable data, therefore try testing your perceptron code with a multitude of data after it works with the OR gate.

---

Solutions

or.pda (26 bytes)	Download PDA32 and the data file below. Place the data file in the same directory as PDA32, and run the program. Then select "or.pda" from the data combo box. The data will be graphed, then press Run to run the perceptron learning process.
cf-ai.zip (14 Kb)	Craig Furness submitted this solution. His neural network tackles the harder problem of the XOR logic gate.