← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9256.d · 生成模型:GPT-6

分段线性激活也可形成非线性

理解等级:中级

ReLU=max(0,z)在两侧分段线性而整体非线性,负半轴梯度为零。Leaky ReLU将负输入映为αz、非负输入保持z;原文把被缩放的一侧写成non negative,应按公式纠正。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      Rectified linear unit (ReLU): The ReLU activation function, similar to the tanh and sigmoid activation functions, consists of a linear and nonlinear region. The ReLU function is given by f (y) = max(0,y). That is, for any negative input, the ReLU outputs the value 0, and for all other values, the ReLU returns in the input value as the output. The ReLU is the most

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第08章《深度学习模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/21_chapter08.xhtml) - 结论依据:摘录中关于分段线性激活也可形成非线性的定义、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:Rectified linear unit (ReLU): The ReLU activation function, similar to the tanh and sigmoid activation functions, consists of a linear and nonlinear region. The ReLU function is given by f (y) = max(0,y). That is, for any negative input, the ReLU outputs the value 0, and for all other values, the ReLU returns in the input value as the output. The ReLU is the most