In MATLAB we generally give Inputs and Target output values corresponding to target input values. For example, I have given values Load, Speed as Input and NOx, Smoke as output. Input Speed is 1200, 1400, 1600, 1800 and 2000 rpm and Load is 10, 20, 30 and 40 Nm. After training, I need the NOx and Smoke values at Load 12 Nm and speed 1708 rpm. How to find out? Neural Networks in MATLAB is perplexing, every paper explains that they give input, output and get R values graph. How to use the training further to get values at desired input points?
So . . . . I think you may not yet have the necessary fundamental understanding of how neural networks operate. Effectively (like all models), they map some space so that when you provide certain values (the input), they output other corresponding values (the output). Training (all models) is merely what makes them map the space accurately. You don't "use the training further to get values at desired input points". You simply input "Load 12 Nm and speed 1708 rpm" into the trained model and it will output the correct "NOx and Smoke values".
MATLAB is not an easy tool to learn data science (though it was AWESOME when it was all we had). I would highly recommend the (free) Azure Machine Learning Studio (https://studio.azureml.net/) to get up to speed before returning to MATLAB (which I assume is required for your courses). Do the initial tutorial (it uses a model other than neural networks but they work exactly the same way) to understand the distinction between training and using the trained model (the web service) to get your needed answers.
(up-votes appreciated)
https://studio.azureml.net/
net = fitnet(10);
net = train(net,x,t);
%To see the outputs for training data
net(x)
%To get output for a sample
q=[12;1708]
net(q)
after training of ANN, you got trained network. transfer that net into nntool and transfer also import input by using import option and simulate it by using nntool.
- Badges
- Science topic