基于自归一化神经网络的脉冲星候选体选择  

Pulsar candidate selection based on self-normalizing neural networks

作　　者：康志伟 刘拓 刘劲[2] 马辛[3] 陈晓[4] Kang Zhi-Wei;Liu Tuo;Liu Jin;Ma Xin;Chen Xiao(College of Computer Science and Electronic Engineering,Hunan University,Changsha 410082,China;College of Information Science and Engineering,Wuhan University of Science and Technology,Wuhan 430081,China;College of Instrument Science and Opto Electronic Engineering,Beihang University,Beijing 100191,China;Shanghai Institution of Satellite Engineering,Shanghai 200240,China)

机构地区：[1]湖南大学信息科学与工程学院,长沙410082 [2]武汉科技大学信息科学与工程学院,武汉430081 [3]北京航空航天大学仪器科学与光电工程学院,北京100191 [4]上海卫星工程研究所,上海200240

出　　处：《物理学报》2020年第6期276-283,共8页Acta Physica Sinica

基　　金：国家自然科学基金(批准号:61772187,61873196)资助的课题~~

摘　　要：脉冲星候选体选择是脉冲星搜寻任务中的重要步骤.为了提高脉冲星候选体选择的准确率,提出了一种基于自归一化神经网络的候选体选择方法.该方法采用自归一化神经网络、遗传算法、合成少数类过采样这三种技术提升对脉冲星候选体的筛选能力.利用自归一化神经网络的自归一化性质克服了深层神经网络训练中梯度消失和爆炸的问题,大大加快了训练速度.为了消除样本数据的冗余性,利用遗传算法对脉冲星候选体的样本特征进行选择,得到了最优特征子集.针对数据中真实脉冲星样本数极少带来的严重类不平衡性,采用合成少数类过采样技术生成脉冲星候选体样本,降低了类不平衡率.以分类精度为评价指标,在3个脉冲星候选体数据集上的实验结果表明,本文提出的方法能有效提升脉冲星候选体选择的性能.Pulsar candidate selection is an important step in the search task of pulsars.The traditional candidate selection is heavily dependent on human inspection.However,the human inspection is a subjective,time consuming,and error-prone process.A modern radio telescopes pulsar survey project can produce totally millions of candidates,so the manual selection becomes extremely difficult and inefficient due to a large number of candidates.Therefore,this study focuses on machine learning developed in recent years.In order to improve the efficiency of pulsar candidate selection,we propose a candidate selection method based on self-normalizing neural networks.This method uses three techniques:self-normalizing neural networks,genetic algorithm and synthetic minority over-sampling technique.The self-normalizing neural networks can improve the identification accuracy by applying deep neural networks to pulsar candidate selection.At the same time,it solves the problem of gradient disappearance and explosion in the training process of deep neural networks by using its self-normalizing property,which greatly accelerates the training process.In addition,in order to eliminate the redundancy of the sample data,we use genetic algorithm to choose sample features of pulsar candidates.The genetic algorithm for feature selection can be summarized into three steps:initializing population,assessing population fitness,and generating new populations.Decoding the individual with the largest fitness value in the last generation population,we can obtain the best subset of features.Due to radio frequency interference or noise,there are a large number of non-pulsar signals in candidates,and only a few real pulsar signals exist there.Aiming at solving the severe class imbalance problem,we use the synthetic minority over-sampling technique to increase the pulsar candidates(minority class)and reduce the imbalance degree of data.By using k-nearest neighbor and linear interpolation to insert a new sample between two minority classes of samples that are

关 键 词：脉冲星候选体选择 自归一化神经网络 特征选择 类不平衡 

分 类 号：P145.6[天文地球—天体物理] TP183[天文地球—天文学]