知识卡片
多类交叉熵等于真实类负对数
独热互斥标签下CE=−log p_c,若p=softmax(z),可写−z_c+logΣexp(z)。z是未归一logit,p才是概率;原书10.21把预测概率与logit记号混用,直接再对概率做Softmax会改变训练目标。
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
- · 人工神经网络
原书出处 · 1 处
- · 查看原书依据
with c the true class (the other terms are zero). Combined with the softmax activation function, the categorical cross-entropy therefore computes as 10.6 Convolutional Neural Networks A drawback of the ANN architectures that we have covered so far is that they do not consider the spatial information that might be hidden in a dataset. For example, as detailed in chapter 8, in the context of
参考来源
- 位置:《Introduction to Autonomous Robots Mechanisms, Sensors, Actuators, and Algorithms (Nikolaus Correll, Bradley Hayes etc.)》第10章《人工神经网络》(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/chapter_10.xhtml)
- 结论依据:摘录中关于多类交叉熵等于真实类负对数的定义、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:with c the true class (the other terms are zero). Combined with the softmax activation function, the categorical cross-entropy therefore computes as 10.6 Convolutional Neural Networks A drawback of the ANN architectures that we have covered so far is that they do not consider the spatial information that might be hidden in a dataset. For example, as detailed in chapter 8, in the context of