Help with simple ANN for data prediction

I am a mechanical engineering student and this is my first time venturing into coding. I wish to create a Neural network for predicting an experimental output.

I have three inputs Let's say Power(P), feed(F), speed(V) and Two outputs Let's say X, Y

I hear that there different types of Neural networks like ANN, RNN and CNN etc. These are some which I know of, so I want to create a customised Neural network in which we can change the no. of neurons, layers and training functions and different aspects to compare different networks and decide the best one with higher prediction accuracy.

Can anyone please help.

Answer 1

[Athul Prakash]

If you're looking to pick up how Neural Networks are handled in Matlab, I would suggest going through some of the examples in the documentation. There are multiple how-to's available for training simple ANNs, CNNs etc on relatively simple datasets.

Matlab supports various several common deep learning layers. You may build a deep learning network by creating an array of adjacent layers. Have a look at the following doc on layers (go through the examples):

https://in.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.layer.html

You may stack FullyConnectedLayers to build an ANN architecture, or include Convolutional layers (such as convolution2DLayer) to build a CNN.

trainNetwork() is invoked to train a DL network on a given dataset and predict(), confusionmat() are useful to evaluate its performance.

You may also find analyzeNetwork() useful.

Hope it helps.

[RAVI Chandan]

Thank you