知识卡片
隐藏层误差累加全部下游路径
隐藏节点影响下一层多个节点,其δ为自身激活导数乘各下游δ按连接权重加权之和。遗漏一个分支会少算梯度;共享参数还需汇总全部使用位置的贡献,而非只取最后一次路径。
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
This allows us to provide the error at node j in layer k, also known as the backpropagation formula: With this last part, we are able to define a recursive definition to calculate the desired error gradient with respect to all weights in the neural network: This computation can be executed layer by layer, starting from the output layer and working its way backward. This
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:摘录中关于隐藏层误差累加全部下游路径的定义、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:This allows us to provide the error at node j in layer k, also known as the backpropagation formula: With this last part, we are able to define a recursive definition to calculate the desired error gradient with respect to all weights in the neural network: This computation can be executed layer by layer, starting from the output layer and working its way backward. This