知识卡片
核心张量连接不同潜在轴
Tucker形式用每个轴的因子矩阵表示较低维特征,再通过核心张量表达这些特征之间的交互。核心连接方式比只配对同号因子更灵活,但参数与估计负担也随核心大小变化。表示结构本身不保证推荐更好。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
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
A core tensor is generated to make those matrices interact with each other. Then, a new third-order tensor is formed by the product of the core tensor and factor matrices. The model is trained by generating an objective function, and every objective function brings a loss function and the regularization associated with it to avoid over-fitting. The algorithm iterates the factor matrices and the core
参考来源
- 位置:《Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications》第11章《张量分解的兴趣点推荐》(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-C11.html)
- 结论依据:摘录中关于核心张量连接不同潜在轴的定义、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:A core tensor is generated to make those matrices interact with each other. Then, a new third-order tensor is formed by the product of the core tensor and factor matrices. The model is trained by generating an objective function, and every objective function brings a loss function and the regularization associated with it to avoid over-fitting. The algorithm iterates the factor matrices and the core