← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8863.a · 生成模型:GPT-6

固定网络也能优化输入表示

理解等级:中级

将层写为xₗ₊₁=fₗ₊₁(xₗ,θₗ₊₁),可选择固定θ而优化输入或某个激活,以实现可视化、提示或其他目标。可求输入导数不意味着任意隐藏激活都可由真实输入实现,也不意味着优化激活与训练权重具有相同约束。

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

      hold the parameters fixed and learn the values of the activations that achieve some objective. In fact, this is what is done in many methods such as network prompting, adversarial attacks, and network visualization, which we will see in more detail in later chapters.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第12章《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/c012.xhtml) - 结论依据:摘录中关于固定网络也能优化输入表示的定义、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:hold the parameters fixed and learn the values of the activations that achieve some objective. In fact, this is what is done in many methods such as network prompting, adversarial attacks, and network visualization, which we will see in more detail in later chapters.