Develop an MLP application system that can deal with .cal data set.
- Analyze the data structure requirements of an MLP to design a class for the MLP neural network system with the following capabilities:
- can read in a .cal data set file.
- can configure an NN based on user’s specification of hidden neurons and the read-in data set.
- can normalize values of each data field of the training and testing data sets.
- can randomly shuffle the data instances in the data sets to generate different sets of training data and testing data.
- can perform an epoch of data training and report the root mean square of the error.
- can test the trained NN using the testing data and report the correctness based on the classification confusing table.
- can perform a simple forward computation using raw input vector (normalized by your code) and return raw output vector (converting back).
- Add graphics display for user to visualize the structure of the NN and the process of training.
- Prepare a folder named as <your ID><your name>Ass12 to put your source code in it. Compress it as an rar file; submit the rar file to course web site.
Appendix: sample code snippets
