Lasso Regression with Quantum Whale Optimization Algorithm

摘要

As the result of regularization of the objective function of linear regression, lasso regression is a classical algorithm of supervised learning in machine learning, and it has a wide range of applications. However, its objective function has the defect of poor derivability, so it does not use the coordinate descent method, but the traditional solution method is the coordinate descent method. But even if the poor derivability is avoided, the method of descending along the coordinate axis also has some defects. For example, when the lasso is more complex, it will obviously reduce the speed; it is easy to fall into the local optimization. In order to solve these defects, this paper chooses a non-convex quantum whale optimization algorithm which is processed by quantum algorithm and has good parallelism on the basis of whale optimization algorithm.