MSE value of neural network model too large

Hello I have a question I need to model a neural network composed of 2 inputs and 7 outputs with 900-size data , when I use the nnstart tool the model is calculated with a single iteration with zero calculation time and gives me values of MSE too large, what should I do to fix this problem? Thank you so much for your help

Answer 1

[TARUN]

Hi @cpge agadir,

I understand that you are facing large MSE error in your neural network.

There can be several reasons that could lead to a high MSE like outliers, missing data, high learning rate and number of iterations, details of which are discussed below.

Here are some of the ways that you can consider to reduce the MSE:

1. Preprocessing of data: Normalize or standardize your input data so that all inputs are on a similar scale. Also, look for any missing values or outliers present.
2. Learning Rate: A high learning rate can cause the model to converge too quickly to a suboptimal solution.
3. Iterations: Increase the number of iterations for better results.
4. Training Algorithm: Try more advanced algorithms like Adam or RMSProp.

You can learn more about the Adam and RMSProp algorithm here:

Adam algorithm: https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.trainingoptionsadam.html

RMSProp: https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.trainingoptionsrmsprop.html

These are some steps that can help you reduce the Mean squared error of the neural network.