AdaBoost integrated learning method based on different-element learners

Abstract

The disclosure provides an AdaBoost integrated learning method based on the different-element learners. The method comprises steps of preprocessing a training set, and initializing the weight of the training set; setting the number of iterations and a threshold value of an AdaBoost algorithm; inputting the training set into an AdaBoost algorithm library, and calculating to obtain an error rate of each different-element learner; selecting the meta learner corresponding to the algorithm with the lowest error rate as a current basic model, and calculating the weight of the current basic model; adjusting the weight of the training set according to the weight of the current basic model to obtain a new training set; and iterating the process, and when the number of iterations is less than or equal to a set number of iterations or the error rate is less than or equal to a set threshold value, performing weighted combination on the different-element learners obtained in the iteration process to obtain an AdaBoost algorithm model. The method has the advantages of strong generalization ability, high calculation efficiency and relatively high accuracy, gives full play to the advantages of each element learner model, avoids the defect of overfitting of a single learner, and can adapt to real-time threat assessment under a battlefield high-dynamic change situation.

Description

technical field [0001] The disclosure belongs to the technical field of computer software, and in particular relates to an AdaBoost integrated learning method based on various meta-learners. Background technique [0002] With the rise of artificial intelligence, unmanned weapons will play a decisive role in the future battlefield, and it is preemptive to make a scientific and reasonable assessment of the current situation of the battlefield and comprehensively consider the complex battlefield situation under the influence of multiple factors on the battlefield. , The key to winning the initiative on the battlefield. For the cluster confrontation scenario between multiple unmanned platforms, the internal target of the cluster needs to estimate the threat degree of the target after obtaining the battlefield situation and environment, and quantitatively describe the threat degree of the target to our multiple unmanned platforms. During the confrontation process, threat estimat...

AI Technical Summary

Problems solved by technology

Although the neural network algorithm is intelligent enough to adapt to changes in the battlefield situation by updating weights and thresholds according to changes in the battlefield situation, it is difficult to set the structure of the network, the interpretability of the algorithm is weak, and the algorithm is time-sensitive. low sex issues

Images