BFELFISH的博客

What is machine learning 目前尚未有统一 的定义 Arthur Samuel对机器学习的定义：Filed of study that gives computers the ability to learn without being explicitly programmed Tom Mitchell对机器学习的定义：Well-posed Learning P...

1.supervised learning监督学习：we gave the algorithm a data set，in which the right answers were give. And the task of the algorithm was to just produce more of these right answers.我们给一个包含正确答案的数据集给算法，数据集的...

1.训练样本大量的情况下，计算偏导数的计算量过大解决：随机梯度下降将从样本中抽取部分样本分成若干份，然后做梯度下降。有利于放到GPU中做并行运算，以牺牲精确度换取时间。2.梯度消失问题（梯度弥散问题）激励函数在某一段的导数过小，导致Loss过早的不再下降，精确度过早的不再提高。解决：消除链式法则中发生的连乘式每一项绝对值小于1的情况。1.初始化一个合适的w有可...