Dear Expert,
I used the Neural Network in MATLAB using inputs [10*3] target data [10*1] and one Hidden layer [25 neurons]. then How can I create an equation that correctly estimates the predicted target??
(Based on the ANN created, weights, biases, and related inputs)
Is there a method, tool, or idea to solve this issue and create one final equation that predicts the output?
Not sure I understood the question as I think you asking how to do:
output = model(input)
Anyway, some examples of how to do it in a simple way:
https://www.mathworks.com/help/thingspeak/create-and-train-a-feedforward-neural-network.html
https://www.mathworks.com/help/deeplearning/ug/create-simple-deep-learning-network-for-classification.html
And if you want to extract network it is more complicated as you would need to extract weights, perform vector multiplications and activation functions computations like:
https://towardsdatascience.com/the-heart-of-artificial-neural-networks-26627e8c03ba
Thank you for your reply, but this is not my equation, I want to generate the final correlation (equation) Just like Y= mX+c where m, c is my constant Y (output) is the dependent variable and X is the independent variable (input), so that we can predict the value for new input.
could you please tell me how i can get the correlation (Equation) from ANN.
Dear Chhotelal Prajapati ,
To create an equation based on an ANN, you will need to specify the input variables and the desired output, and then design and train the ANN to learn the relationship between the inputs and the output.
https://www.mdpi.com/2076-3417/13/14/8078
Regards,
Shafagat
Just as previous comment said, you need to write down relationships between nodes. ANN is a complex collection of funtions so simple correlation does not exist. However, you can write it down as function and use some tools or do simplification by hand, it is all based on number of layers your nework has. Yet, looking at input size - it's not going to be easy.
Dear Andrius Ambrutis
Yes, you are right, but the things I want to know is how can I generate the equation which gives output based on the number of input variables and constants. Can you tell me what tools, ideas and method I need to follow so that I can generate the equation when the number of layer is more than one and neuron is also more than 2 because it is not mathematically easy to solve by hand , I need to write code or script or function to create the equation, can you please suggest the answer.
You should read articles about gray-box neural networks. We take that ANN is black-box model but researchers stated that by doing some transformations we can make ANN to become gray-box model. In a way that might be what can help you to simplify model and make it understandable. I only know some add-ons which can convert ANN to formula but not simplify it. If activation functions are the same it shouldn't be too hard to simplify network as:
https://alasko.medium.com/convert-simple-neuron-network-to-mathematician-notation-58a0d72f0337
Problem is if you use multiple activation functions.
Well, if you use ReLU which is max(0, x) it even should be easy to make simple equation or... You can use linear activation function, which is kinda no function at all. I never tried it and it sounds a bit stupid but netword should still work well for some tasks. People use it for output but you might find some research which use it for network itself.
https://www.v7labs.com/blog/neural-networks-activation-functions#:~:text=The%20linear%20activation%20function%2C%20also,the%20value%20it%20was%20given.
Not sure if I can help with anything more.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts