基于鲸鱼优化和深度学习的不平衡大数据分类算法

Unbalanced Big Data Classification Algorithm Based on Whale Optimization and Deep Learning

针对当前不平衡数据分类算法中存在的分类精度低和容易陷入局部最优状态的问题,提出一种基于鲸鱼优化和深度学习的不平衡大数据分类算法.该算法由特征选择、预处理和分类3个阶段组成:①为了提高分类精度,使用鲸鱼优化算法(whale optimization algorithm,WOA)在不平衡数据中寻找最优特征子集,消除不相关和多余的特征;②采用局部敏感哈希的合成少数类过采样技术(locality sensitive hashing synthetic minority oversampling technique,LSH-SMOT)对数据集进行预处理,解决类不平衡问题;③使用基于WOA算法优化的双向递归神经网络(bidirectional recurrent neural networks,BRNN)对预处理后的数据集进行分类.实验结果表明:本文算法能够有效解决不平衡数据集的分类问题,相比于其他算法,本文算法在分类精度和局部最优避免率方面具有明显优势.

Aiming at the problems of low classification accuracy,which is easy to fall into local optimal state in the current unbalanced data classification algorithm,an unbalanced big data classification algorithm based on whale optimization and deep learning has been proposed.The algorithm consists of three parts:feature selection,preprocessing and classification.Firstly,in order to improve the classification accuracy,the whale optimization algorithm(WOA)has been used to find the optimal feature subset in the unbalanced data to eliminate the irrelevant and redundant features.Secondly,the locality sensitive hashing synthetic minority oversampling technique(LSH-SMOTE)has been used to preprocess the dataset to solve the class imbalance problem.And,finally,the bidirectional recurrent neural networks(BRNN)optimized based on WOA algorithm been used to classify the preprocessed dataset.The experimental results show that the proposed algorithm can effectively solve the classification problem of unbalanced data sets.Compared with other algorithms,it has obvious advantages in classification accuracy and local optimal avoidance rate.