This paper proposes a new fast algorithm for solvingleast squares kernel twin bounded SVR .Different from classical algorithm ,the fast algorithm can obtain the regression function by using the KKT conditions of the original problem instead of solving the dual problem .In order to verify the ef‐fectiveness of the fast algorithm ,in this paper ,a series of experiments compared with the classical algo‐rithm are performed by using 10 datasets taken from UCI database and 4 evaluating indicators .The ex‐periment results indicates that the proposed fast algorithm is an effective and competitive algorithm .%提出一个新的求解最小二乘核双生有界SVR的快速算法。与经典算法不同的是,快速算法不是通过求解对偶问题,而是通过原始问题的KKT 条件得到回归函数。为了验证快速算法的有效性,本文利用U C I数据库中的10个数据集和4个评价指标与经典算法进行了一系列的比较实验。实验结果表明所提算法是一个有效的,可竞争的算法。
展开▼