论文部分内容阅读
分析在特定假设空间下k-部排序学习算法的可学习性.给出k-部排序可学习和可有效学习的概念,得到样本复杂度的上界以及k-部排序算法可有效学习的一个充分条件,同时给出与计算复杂度相关的若干结果.最后,将部分结果推广到限制模型中.
We analyze the learning ability of k-order learning algorithm in a given hypothesis space.We give the concept of k-ordered rank learning and efficient learning, get the upper bound of the sample complexity, and the k-rank ordering algorithm can effectively learn one Sufficient conditions are given, and some results related to computational complexity are given.Finally, some results are generalized to the restricted model.