The Cobb-Douglas Learning Machine
作者：
Highlights：
• A novel Minimum Error Minimax Probability Machine (MEMPM) method is presented.
• The Cobb-Douglas production function is extended to machine learning.
• The proposal is a robust formulation for linear and kernel-based classification.
• The method is solved via a self-developed two-step alternating algorithm.
• We prove that the optimization scheme converges to the optimal solution of the problem.
• Best performance is achieved in experiments carried out on 17 benchmark datasets.
摘要
•A novel Minimum Error Minimax Probability Machine (MEMPM) method is presented.•The Cobb-Douglas production function is extended to machine learning.•The proposal is a robust formulation for linear and kernel-based classification.•The method is solved via a self-developed two-step alternating algorithm.•We prove that the optimization scheme converges to the optimal solution of the problem.•Best performance is achieved in experiments carried out on 17 benchmark datasets.
论文关键词：Cobb-Douglas,Minimax Probability Machine,Minimum Error Minimax Probability Machine,Second-order Cone Programming,Support Vector Machines
论文评审过程：Received 4 January 2021, Revised 10 March 2022, Accepted 5 April 2022, Available online 7 April 2022, Version of Record 15 April 2022.
论文官网地址：https://doi.org/10.1016/j.patcog.2022.108701
