摘要
神经网络算法在数据分类与计算中对非对称数据的处理是一项亟待解决的难题。文章提出一种修改权重的逆传输类神经网络算法,通过修改自学习效率,对占有较少类的数据分配高权重来解决非对称平衡问题。仿真结果表明,与其他五种分类算法对比,本算法在不影响算法复杂度的情况下,提高了对非对称数据运算的精确性与有效性。
Neural network algorithm in data classification and calculation of asymmetric data processing is an urgent problem.Therefore,this paper proposes a modification of the weights inverse transmission neural network algorithm.By modifying the efficiency of self-learning and assigning high weights to data that occupies fewer classes,the asymmetric balance problem is solved.Compared with the other five classification algorithms by simulation data,this algorithm does not affect the algorithm complexity of the situation,improving the operation of asymmetric data accuracy and effectiveness.
作者
魏传佳
WEI Chuan-jia(Quanzhou College of Technology,Quanzhou,Fujian,China 362200)
出处
《湖南邮电职业技术学院学报》
2021年第3期22-25,58,共5页
Journal of Hunan Post and Telecommunication College
基金
福建省教育厅2020年科研项目“基于模拟退火算法的无线网络优化算法研究”(项目编号:JAT201502)。
关键词
神经网络
非对称数据
逆向传输
算法有效性
neural network
asymmetric data
reverse transmission
effectiveness of the algorithm