知识卡片
随机梯度来自抽样估计
用随机样本或小批估计全训练目标梯度,可降低每步内存与计算,并带来噪声。仅重排全批所有项再求同一个和,不会产生同等随机更新效果;噪声可能帮助探索,也不保证逃离所有局部极小。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 神经网络基础
原书出处 · 1 处
- · 查看原书依据
Up until now, you have been aggregating the error for all the training examples and skiing down the steepest route as fast as you can. But training on the entire training set, one sample at a time, is a little nearsighted. It’s like choosing the downhill sections of a snow park and ignoring all the jumps. Sometimes, a good ski jump can help you skip
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第05章《神经网络基础》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-5.html)
- 结论依据:摘录中关于随机梯度来自抽样估计的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Up until now, you have been aggregating the error for all the training examples and skiing down the steepest route as fast as you can. But training on the entire training set, one sample at a time, is a little nearsighted. It’s like choosing the downhill sections of a snow park and ignoring all the jumps. Sometimes, a good ski jump can help you skip