摘要
非无限步下波动数据的数学分类问题是否具有有限收敛性是解决数学问题分类的关键。对该类数学问题进行了建模,验证了非无限步下波动数学分类问题具有可行性和有限收敛性。通过凸优化KKT等价条件,从边缘衰减不能为空、矩阵更新过程中存在正确策略和可逆、波动分类集边缘过程中不会移出集合这3方面对模型的可行性和有限收敛性进行验证。仿真实验对3类实际采集数据集进行模型验证,结果表明模型具有可行性和有限收敛性。
The infinite fluctuations on whether mathematics classification problem has finite convergence is the key to solv-ing math problems. In this paper, the mathematical modeling issues, and to verify the mathematical classification problem of the infinite on volatility has finite convergence. Convex optimization equivalent KKT conditions, and from the edge of the attenuation can＆#39;t be empty, right exist in the process of matrix updating strategy and reversible, fluctuations in question classification to set the edge process won＆#39;t be removed from the collection of the three parties face model and the feasibility of the finite convergence for validation. Simulation model validation was carried out on the three kinds of actual data sets, the results show that the model is feasibility and convergence.
出处
《科技通报》
北大核心
2014年第6期7-9,共3页
Bulletin of Science and Technology
基金
四川省科技项目(20114A369)
关键词
非无线步下波动
数字分类
矩阵
有限收敛性
infinite fluctuation
mathematics classification
matrix
finite convergence
