一种新的多任务朴素贝叶斯学习方法

New Implementation of Multi-Task Na■ve Bayes Learning

针对现有多任务学习方法运行效率低和数据信息利用不足的问题,提出了一种新的多任务朴素贝叶斯学习方法。该方法基于朴素贝叶斯原理,将经典的单任务朴素贝叶斯模型引入到多任务学习问题中,通过两种新的更新策略,更好地利用总体数据的先验信息来提高模型的分类性能和泛化能力;该方法与目前的多任务学习模型相比,运行效率显著提升。实验结果表明,该方法优于只针对单任务的朴素贝叶斯学习方法,并在部分数据集上优于当前主流的多任务学习方法。

The multi-task learning models often exist the problems of low efficiency and insufficient use of shared data information,this paper proposed a new implementation of multi-task Na■ve Bayes learning method. Based on the Na■ve Bayesian principle,this method introduces the classical singletask Na■ve Bayesian learning model into multi-task learning problem. Through two new updating strategies,the prior information of the overall data is better utilized to improve the classification performance and generalization ability of the model. Compared with the current multi-task learning model,the method has significantly improved the operating efficiency of the current model. The experimental results show that this method is superior to the single-task Na■ve Bayes learning method and is superior to current state-of-the-art multi-task learning methods in some datasets.