知识卡片
输出层误差由损失与激活共同决定
对单输出半平方损失E=½(ŷ−y)²,输出δ=(ŷ−y)g′(a)。换成交叉熵或其他输出映射时需重新求导,不能照搬此式;全批次的平均系数也应同损失定义一致。
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
- · 附录D:反向传播
原书出处 · 1 处
- · 查看原书依据
Together, these two result in We continue to use the chain rule to work backward along the computation graph. Specifically, the activation at node j in layer k, with 1 ≤ k < m, feeds into all nodes l = 1..rk+1 of layer k + 1. Therefore, the error is calculated as Using , the above equation simplifies to Inspecting the computation graph or the
参考来源
- 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》附录D《附录D:反向传播》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)/_epub-src/OEBPS/xhtml/appendixD.xhtml)
- 结论依据:摘录中关于输出层误差由损失与激活共同决定的定义、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:Together, these two result in We continue to use the chain rule to work backward along the computation graph. Specifically, the activation at node j in layer k, with 1 ≤ k < m, feeds into all nodes l = 1..rk+1 of layer k + 1. Therefore, the error is calculated as Using , the above equation simplifies to Inspecting the computation graph or the