Increasing system performance in machine learning by using multiprocessing

Machine Learning regained its popularity after the recent advances occurred in deep learning area and Python programming language is increasingly being used in machine learning. It is important to choose proper initial values for variables called hyper parameters used in machine learning. Selecting proper initial values for these variables directly effects the systems performance at the training, testing and validation steps of learning model. In some situations optimal initial values for hyper parameters are found by trial method. This means that so many trials may need to be performed and these trials consumes much lower time once the computers multiprocessing feature is used. In this work, it is shown how some operations which may be needed in machine learning methods are performed on more than one processor by using simple scripts written in Python programming language. Moreover the time gain obtained with these scripts is calculated.