Constructing neural netowrk (perceptron)

[nine]

Homework Statement

Implement (AND, NAND) gates using MATLAB.

Homework Equations

The Attempt at a Solution

I've succeeded to implement a single perceptron that has a single output and multi inputs.
Here is the code I've written
--------------------------------------------------------------------------
P = [0 0 1 1;0 1 0 1] %Pattern
T = [0 0 0 1] %Target
net = newp(minmax(P),1)
net = train(net,P,T)
--------------------------------------------------------------------------
I need two output terminals, AND results goes to the first terminal and the inverted AND (NAND goes to the other terminal).
I've referred to MATLAB help



Description


train trains a network net according to net.trainFcn and net.trainParam.

train(net,P,T,Pi,Ai,VV,TV) takes

netNetworkPNetwork inputsTNetwork targets (default = zeros)PiInitial input delay conditions (default = zeros)AiInitial layer delay conditions (default = zeros)VVStructure of validation vectors (default = [])TVStructure of test vectors (default = [])


When I tried to rewrite my code with a 2 row target matrix I've got an error as follows

P = [0 0 1 1;0 1 0 1] %Pattern
T = [0 0 0 1;1 1 1 0] %Target
net = newp(minmax(P),1)
net = train(net,P,T)

? Error using ==> network.train
Targets are incorrectly sized for network.
Matrix must have 1 rows.

Error in ==> AndGate at 4
net = train(net,P,T)

My question is, how can I set two output terminals instead of single output terminal? as I need one terminal for AND gate and the other for NAND gate.

Thanks.

 

[KataKoniK]

Hello, to implement both AND and NAND gates using MATLAB, you will need to create a network with two output neurons instead of one. This can be done by changing the number of neurons in the output layer when creating the network using the newp function. For example:

net = newp([0 1; 0 1],2)

This will create a network with two inputs (0 and 1) and two output neurons. Then, you can train the network using the train function, passing in the target matrix T with two rows (one for AND and one for NAND). Finally, you can use the sim function to simulate the network and get the outputs for both gates. For example:

output = sim(net,P)

This will give you a 2x4 matrix, with the first row representing the output for AND and the second row representing the output for NAND. You can then use these outputs to test the accuracy of your network. Hope this helps!

 

[piercas]

Hello,

To construct a neural network with multiple output terminals, you will need to use a different type of network called a multi-layer perceptron (MLP). This type of network allows for multiple outputs and can be trained to perform different tasks simultaneously.

To implement the AND and NAND gates using MATLAB, you can use the "newff" function to create an MLP with two output neurons. The code would look something like this:

P = [0 0 1 1;0 1 0 1]; %Pattern
T = [0 0 0 1;1 1 1 0]; %Target
net = newff(minmax(P),[2 2]); %Create MLP with 2 input neurons and 2 output neurons
net.trainFcn = 'trainlm'; %Set training function to Levenberg-Marquardt
net = train(net,P,T); %Train the network

You can then use the "sim" function to test the network on new inputs and see the outputs for both the AND and NAND gates.

I hope this helps. Good luck with your project!

 

[piercas]

Dear student,

Great job on successfully implementing a single perceptron with multiple inputs! To create a perceptron with two output terminals, you will need to modify your code to create a network with two output neurons instead of one. You can do this by changing the number of output neurons in the newp function:

net = newp(minmax(P),2)

This will create a network with two output neurons, which you can then train using the train function. You will also need to modify your target matrix to have two rows, with the first row representing the AND gate results and the second row representing the NAND gate results:

T = [0 0 0 1;1 1 1 0]

Once you have trained your network, you can use the sim function to test the performance of your network on new inputs.

I hope this helps. Keep up the good work in your neural network studies!