Random projection is a well-known AI calculation, which can be executed by neural organizations and trained in an effective way. Adaptive Regularize Parameter Selection it will regularize the features which are important and select a most suitable feature for a specific problem. The main goal is to reduce the computational cost of both classification and regression task by using Randomization Algorithm. It provides the best possible result to the high-dimensional optimization. It uses Multi layer neural network for performing linear and non linear function.