Fast and efficient kernel machines using random kitchen sink and ensemble methods

ABSTRACT: The introduction of the information era has affected all areas of computer science. This has also affected the approach taken by machine learning researchers. Compared to the early days, researchers now use a generic approach to selecting a machine learning model. As the generic model doesn’t contain much domain knowledge, it has been compensated for by a huge dataset. This makes the model optimization more complex and time consuming. We can replace this overhead with randomization instead of optimization. There are many existing methods which apply randomization in machine learning models. But all of these methods compromise on accuracy. This paper shows an efficient way to use randomization along with ensemble methods without a decrease in efficiency.