Logistic Regression Based on Halton Sequence Initialized Lion Swarm Optimization Algorithm

Abstract

Logistic regression is a classical nonlinear binary classification algorithm with supervised learning in the field of artificial intelligence (AI). Its idea is to get a logical regression after the nonlinearization of linear regression through Sigmoid firstly, and then to get the optimization objective function - cross entropy loss function. Generally, it is solved by gradient descent (GD) or stochastic gradient descent (SGD) method, but these methods are easy to fall into the trap of local minimum; moreover, the closer it is to the optimal value, the sawtooth effect is easy to appear, resulting in the reduction of the operation efficiency of the algorithm. Therefore, in this paper, a new derivative-free optimization method - Halton sequence initialized lion swarm optimization algorithm has been proposed to replace the original GD method or SGD method, which makes the logistic regression algorithm more efficient, and via numerical experiment, the effectiveness of the proposed algorithm has been demonstrated.