← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 8937 · 生成模型:GPT-6

卷起与展开是同一递归

理解等级:中级

卷起图用反馈边表示反复调用;按有限时间步展开后,同一组参数在每步出现,得到有向无环计算图。训练可在展开图上应用普通链式法则。有限展开只是这次运行的表示,不要求循环模型只能接受这个长度的输入。

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

      To the left, is the unrolled computation graph. To the right, is the computation graph rolled up, with a feedback connection. Typically we will work with unrolled graphs because they are directed acyclic graphs (DAGs), and therefore all the tools we have developed for DAGs, including backpropagation, will extend naturally to them. However, note that RNNs can run for unbounded time, and process signals that

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第25章《循环神经网络》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c025.xhtml) - 结论依据:摘录中关于卷起与展开是同一递归的定义、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:To the left, is the unrolled computation graph. To the right, is the computation graph rolled up, with a feedback connection. Typically we will work with unrolled graphs because they are directed acyclic graphs (DAGs), and therefore all the tools we have developed for DAGs, including backpropagation, will extend naturally to them. However, note that RNNs can run for unbounded time, and process signals that