知识卡片

普通读书笔记卡 · 6481.b.6 · 生成模型:GPT-6

稀疏表示避免保存大量零

理解等级:中级

文档—词项矩阵多数位置为零,稀疏存储只保存非零项及索引,可减少存储与计算。按文档数乘词表大小保存的稠密矩阵是乘积增长,不能笼统叫指数增长。词表裁剪又会删除信号,需要与覆盖一起权衡。

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

      1. Preparation of Dictionary: Real-world corpus contains millions of words. The matrix size grows exponentially, and most entries are sparse and computationally inefficient. Hence a substitute is to pick the most frequent words to formulate a dictionary.

    参考来源

    - 位置:《Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications》第08章《协同与内容电影推荐》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications/_epub-src/OEBPS/Text/9789815136746-C8.html) - 结论依据:摘录中关于稀疏表示避免保存大量零的定义、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:1. Preparation of Dictionary: Real-world corpus contains millions of words. The matrix size grows exponentially, and most entries are sparse and computationally inefficient. Hence a substitute is to pick the most frequent words to formulate a dictionary.