基于动态步长交替方向乘子法正则化极限学习机OA北大核心CSTPCD
Regularized extreme learning machine based on variable step alternating direction method of multipliers
为解决交替方向乘子法(alternating direction method of multipliers,ADMM)正则化极限学习机(regularized extreme learning machine,RELM)迭代收敛速度慢和迭代后期误差衰减停滞的问题,提出一种基于动态步长ADMM的正则化极限学习机,记为VAR-ADMM-RELM.该算法在ADMM算法的基础上采用动态衰减步长进行迭代,并同时使用L1和L2正则化对模型复杂度进行约束,解…查看全部>>
To address the deficiency of slow convergence rate and stagnation of error decay during later iteration of alternating direction method of multipliers(ADMM)for regularized extreme learning machine(RELM),we propose a dynamic step size ADMM-based RELM algorithm denoted as VAR-ADMM-RELM.This method iterates with dynamically decaying step sizes based on the ADMM algorithm and simultaneously constrains the model complexity using both L1 and L2 regularization,such…查看全部>>
卢辉煌;邹伟东;李钰祥
北京理工大学自动化学院,北京 100081北京理工大学自动化学院,北京 100081北京理工大学自动化学院,北京 100081
计算机与自动化
人工智能机器学习极限学习机交替方向乘子法正则化动态衰减
machine learningextreme learning machinealternating direction method of multipliersregularizationdynamic decay
《深圳大学学报（理工版）》 2024 (3)
264-273,10
National Natural Science Foundation of China(61906015)Natural Science Foundation of Beijing(L201004) 国家自然科学基金资助项目(61906015)北京市自然科学基金资助项目(L201004)
评论