← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6676.a · 生成模型:GPT-6

词表大不自动排除朴素贝叶斯

理解等级:中级

高维、标注少会增加某些模型的估计难度,但词表维数超过样本数并不单独证明朴素贝叶斯不可用。正则、模型假设、特征稀疏性和类别频率都会影响结果;应保留合适基线,不能按维数一项直接指定唯一分类器。

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

      The spaCy tokenizer gave you 19,169 words in your vocabulary. You have almost 4× as many words as you have messages and almost 30× as many words as toxic comments. So your model will not have a lot of information about the words that will indicate whether a comment is toxic or not.

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第04章《从词数到语义分析》(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-4.html) - 结论依据:摘录中关于词表大不自动排除朴素贝叶斯的定义、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:The spaCy tokenizer gave you 19,169 words in your vocabulary. You have almost 4× as many words as you have messages and almost 30× as many words as toxic comments. So your model will not have a lot of information about the words that will indicate whether a comment is toxic or not.