The Application of MD-MIML Frame Based on MIML
In this paper, we presented an improved learning frame based on MIML structure, which is combined with feedback mechanism. In real-world, there are many problems mentioned in complicated systems, traditional multi-instance multi-label structure would only consider the initial and fianl state, when facing with much more complicated system, the lack of middle decision layer affect the whole capability of system. The new proposed structure-MDMIML frame will help us to solve these task more easily by layered MIML with reward machanism. Finally, we use RoboCup Simulation 2D as platform to verify our frame and algorithms, the result shows the frame presented is valid and well-performed.
Multi-instance learning Multi-label learning Multi-instance Multi-label learning D-MIMLSVM
Guo Jing Chen Wei Wang Jie Huang Ying
Faculty of Automation Guangdong University of Technology Guangzhou,China College of Business&Public Administration Governors state University Illinois,USA
国际会议
重庆
英文
126-131
2011-01-21（万方平台首次上网日期，不代表论文的发表时间）