计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2014年
16期
31-34,61
,共5页
韩顺成%马小晴%陈进东%潘丰
韓順成%馬小晴%陳進東%潘豐
한순성%마소청%진진동%반봉
支持向量机回归%序列最小优化算法%对偶间隙%KKT条件%停机准则
支持嚮量機迴歸%序列最小優化算法%對偶間隙%KKT條件%停機準則
지지향량궤회귀%서렬최소우화산법%대우간극%KKT조건%정궤준칙
Support Vector Regression%Sequential Minimal Optimization(SMO)%duality gap%Karush-Kuhn-Tucker (KKT)%Stopping Criteria
在序列最小优化（Sequential Minimal Optimization，SMO）算法训练过程中，采用标准的KKT（Karush-Kuhn-Tucker）条件作为停机准则会导致训练后期速度下降。由最优化理论可知，当对偶间隙为零时，凸二次优化问题同样可以取得全局最优解。因此本文将对偶间隙与标准KKT条件同时作为SMO算法的停机准则，从而提出了改进停机准则的SMO算法。在保证训练精度的情况下，提高了SMO算法的训练速度。通过对一维和二维函数的两个仿真实验，验证了改进SMO算法的有效性。
在序列最小優化（Sequential Minimal Optimization，SMO）算法訓練過程中，採用標準的KKT（Karush-Kuhn-Tucker）條件作為停機準則會導緻訓練後期速度下降。由最優化理論可知，噹對偶間隙為零時，凸二次優化問題同樣可以取得全跼最優解。因此本文將對偶間隙與標準KKT條件同時作為SMO算法的停機準則，從而提齣瞭改進停機準則的SMO算法。在保證訓練精度的情況下，提高瞭SMO算法的訓練速度。通過對一維和二維函數的兩箇倣真實驗，驗證瞭改進SMO算法的有效性。
재서렬최소우화（Sequential Minimal Optimization，SMO）산법훈련과정중，채용표준적KKT（Karush-Kuhn-Tucker）조건작위정궤준칙회도치훈련후기속도하강。유최우화이론가지，당대우간극위령시，철이차우화문제동양가이취득전국최우해。인차본문장대우간극여표준KKT조건동시작위SMO산법적정궤준칙，종이제출료개진정궤준칙적SMO산법。재보증훈련정도적정황하，제고료SMO산법적훈련속도。통과대일유화이유함수적량개방진실험，험증료개진SMO산법적유효성。
In SMO training process, standard KKT stopping criteria will lead to training speed declining along with training progress. According to the optimum theory, if the dual gap is zero, convex quadratic optimization problem will also obtain global optimal solution. Therefore, an improved stopping criteria of SMO is proposed in this paper, it combines the duality gap and standard KKT condition as the stopping criteria. This algorithm can improve the training speed without training accuracy decrease. Two cases experimental simulating results corroborate the efficiency of this algorithm.
