I am a machine learning beginner, recently are learning neural network model and he had to write a multilayer BP neural network, found when using the iris data set for testing, test results is fine for most of the time, for example:

Poor performance but occasionally happen, for example:

This may be because of what reason, please? Writing is my model may have a problem? Or the training set is too small? There is something wrong with the model parameters or me?
Later, I use breast cancer data set tests found that the accuracy is over 95%, basic model of writing should be no problem, I also adjust tried, always can appear sometimes poor performance, don't know why?

CodePudding user response:

Here basically is to want to make sure that even if the model code is still poor performance may come up correctly, if only because there is something wrong with the code that I have a good look at the code, trouble you