← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6685 · 生成模型:GPT-6

表示学习仍需选择数据与目标

理解等级:中级

神经网络可在预测目标反馈下共同学习中间特征,减少逐项手工组合的负担。它仍需要输入表示、架构、损失和训练数据,不保证找到全局最优,也不自动善于少样本或任意未标注任务;评价应看具体基线和独立样本。

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

      Neural networks do the feature engineering for you, and they do it optimally. They extract generally useful features and representations of your data according to whatever problem you set up in your pipeline, and modern neural networks work especially well, even for information-rich data, such as natural language text.

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第05章《神经网络基础》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-5.html) - 结论依据:摘录中关于表示学习仍需选择数据与目标的定义、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:Neural networks do the feature engineering for you, and they do it optimally. They extract generally useful features and representations of your data according to whatever problem you set up in your pipeline, and modern neural networks work especially well, even for information-rich data, such as natural language text.