摘要
极限学习机算法适应新鲜样本能力强、学习速率快,为此提出基于极限学习机算法分析图书馆读者借阅行为。但是极限学习机算法输入权重与隐层阈值随机确定,行为分析结果随机性强、可靠程度低,所以采用高适应度值遗传算法确定极限学习机算法的输入权值与阈值。高适应度值遗传算法选择算子复制两份适应度值最优个体、复制一份适应度值较优个体作为遗传种群;交叉算子选取2个适应度值最优新个体开始变异操作;确定算法最优输入权值与阈值后,提取读者借阅行为特征作为训练样本,构建图书馆读者借阅行为分析模型,测试样本代入模型得到读者借阅行为分析结果。经测试,所提方法能准确分析出高校学生频繁借阅、少量借阅等图书借阅行为。
The extreme learning machine(ELM)algorithm is ofpowerful adaptability to fresh samples and of fast learning rate.Therefore,the borrowing behavior of library readers is analyzed based on the ELM algorithm.However,the input weight and hidden layer threshold of the LEM algorithm are determined randomly,and the behavior analysis results are of serious randomness and low reliability,so the genetic algorithm with high fitness value is adopted to determine the input weight and threshold of the LEM algorithm.The selection operator of the genetic algorithm with high fitness is used to replicate two individuals with optimalfitness value and one individual with better fitness value as the genetic population,and the crossover operator is used to select two new individuals with optimal fitness values to start mutation operation.After determining the optimal input weights and thresholds of the algorithm,the characteristics of readers′borrowing behaviors are extracted as training samples to construct the analysis model of borrowing behaviors.The test samples are subjected to the model to obtain the analysis results of readers′borrowing behaviors.After testing,the proposed method can accurately analyze the borrowing behaviors of college students,e.g.,frequent but few borrowing.
作者
张捷
ZHANG Jie(Langfang Normal University,Langfang 065000,China)
出处
《现代电子技术》
北大核心
2020年第5期121-124,共4页
Modern Electronics Technique
基金
社区图书馆的现状分析及发展对策探析--以廊坊市社区图书馆为例(2015090)。
