Visual Learning by Set Covering Machine with Efficient Feature Selection


   In this paper, we propose a new visual learning method for real-world object recognition task. Our method is based on the Set Covering Machine (SCM), to make the learning time shorter than the methods based on commonly used trial-and-error algorithms, such as genetic programming and reinforcement learning. Generally, the process of visual learning is quite time-consuming because image data consists of large amount of information.
   We attempt to reduce the learning time by introducing the effective feature selection method to find a small number of useful features in image data. Additionally, we introduced a criterion based on the Minimum Description Length (MDL) principle to refine the hypothesis.
   We perform some experiments to verify the effectiveness of our method.



[Back]