论文部分内容阅读
半指导问题是近来机器学习研究中的备受关注一个重要内容．本文以满足“在输入空间中相近的对象其输出也相近”这一源于直观事实的原则(相近原则)去解决半指导学习问题,给出在这个原则下的一个一般的直接推理方法—基于相近原则的半指导问题直推学习机,得到了这个问题的解析解及迭代算法,用模式分类实例验证该方法的有效性,并给出适于在线处理的增量学习算法,这些增量算法尤其还适于新增了有指导的信息的场合．
Semi-coaching is an important issue in the recent research of machine learning. This paper solves the semi-coaching problem with the principle of intuition (similar principle) that “the output of similar objects in input space is similar”, and gives a general direct reasoning method based on this principle - based on Similarity principle of semi-guide direct push learning machine, the analytical solution and iterative algorithm of this problem are obtained, and the validity of this method is verified by the example of pattern classification, and the incremental learning algorithm suitable for on-line processing is given. These incremental algorithms In particular, it is also suitable for the occasion where new guidance information is added.