← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8989.b · 生成模型:GPT-6

潜变量完备依赖生成假设

理解等级:中级

给定确定生成器且将所有随机输入都纳入z,z可以决定生成样本x。但它不必能从任意真实图像唯一恢复,也不保证紧凑、无冗余或有意义。若解码器还从p(x|z)抽样,z本身不包含最终观测噪声的全部信息。

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

      What we did not mention in chapter 32, but will focus on now, is that latent variables are representations of the data. In the case of an unconditional generative model, the latent variables are a complete representation of the data: all information in the data is represented in the latent variables.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第33章《生成建模与表示学习》(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/c033.xhtml) - 结论依据:摘录中关于潜变量完备依赖生成假设的定义、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:What we did not mention in chapter 32, but will focus on now, is that latent variables are representations of the data. In the case of an unconditional generative model, the latent variables are a complete representation of the data: all information in the data is represented in the latent variables.