一种快速判定原始数据集合线性可分的充分条件

摘要

With the deepening of machine learning research , a variety of classification algorithms are widely applied to various fields in real life.However, many classification algorithms often need to determine whether the original data set is linearly separable , so the judgement condition for the separability of original data set has become a research focus .As the existing linear separability judgment algorithms for original data set are with a higher computational complexity , in this paper, we propose a sufficient condition for determining the data linear separability, and on this basis, we give the data separability discriminant algorithm .Theoretical analysis and instance validation all show that the data separability discriminant algorithm proposed in this paper is valid and with a lower computational complexity .%随着机器学习研究的深入，各种各样的分类算法被广泛地应用到现实生活中的各个领域。然而目前的很多分类算法往往需要首先判断原始数据集合是否线性可分，因此关于原始数据集合是否线性可分的判决条件就成了一个研究的热点。针对于现有数据可分性判决算法计算复杂度较高的不足，提出一种判定数据线性可分的充分条件，并在此基础上给出数据可分性的判别算法。理论分析和实例验证说明了提出的数据可分性判别算法是有效的并且具有较低的线性计算复杂度。