How to use trained neural network using new data

Hello esteemed friends, I need help. How can I use neural network to generate an output using 2 or 3 random new input data on trained neural network? For example using the abalone_dataset, I would train the neural network using input data (length, diameter, height, etc.) and the target data. After that, using only 2 to 3 input data (length and height only) to generate an output using the trained neural network.

[Rajeev]

Are you planning to use the Deep Network Designer for it?

[AHMAD FATHURRAHMAN]

I'm still new to the neural network but if the Deep Network Designer can satisfied with the requirements that I need. I really appreciate if you can show me the way.

[Rajeev]

Based on what you have mentioned, you are trying to exploit 'transfer learning'.

You can refer to this link Get Started with Transfer Learning - MATLAB & Simulink (mathworks.com) to know more about it.

If you already have a pre-trained model, you can import it to the Deep Network Designer and modify the classification and output layers to get the output for your data.

Can you share the trained model that you want to modify?

Answer 1

[Varun Sai Alaparthi]

Hello Ahmad,

I understand that you are looking for ways to get output from your trained model by giving random inputs. The function depends on your network type.

If you are using a shallow network, you can directly do

Output = net(X); 

If you are using ‘dlnetwork’ or a network trained using ‘trainNetwork’ you can use the ‘predict’ function to get the output.

Output = predict(net,images); 

If you have any further queries, please feel free to reply to my answer. 

Sincerely,

Varun

[Max Newman]

I trained a signal classifier but when I use predict it says that the "The input images must have a size of [96 64 1]." How can I use a trained network on signal data.