The Weighted Average Ensemble Learning Based on Polar Bear Algorithm with T-Distribution Parameters

Ensemble learning is a supervised classification algorithm in the field of machine learning and artificial intelligence. Its main idea is to combine the weak classifiers into strong classifiers according to a certain strategy, so that ensemble learning becomes a linear weighting function. In this paper, the weighted average ensemble algorithm is used for dichotomy. However, due to the limitation of loss function, the traditional gradient descent method and the random gradient descent method are not available. Therefore, the polar bear algorithm in the swarm intelligence algorithm that does not depend on the gradient information is adopted in this paper to obtain the optimal weight, which overcomes the original shortcomings.