Data reduction for Boolean matrix factorization

In the associated paper, we discuss a fault in a recently published method of data reduction in Boolean matrix factorization. We demonstrate experimentally that reversing the logic of the rationale yields comparable results and, moreover, that a random approach yields comparable and on average even better results as well. This data package contains folders with PDF charts that complement the findings presented in the paper.

在相关论文中，我们探讨了一种近期发布的布尔矩阵降维方法的缺陷。我们通过实验证明，逆转推理的逻辑能够产生可比较的结果，并且更进一步，随机方法产生的结果在平均意义上甚至更为优越。本数据包包含包含PDF图表的文件夹，这些图表补充了论文中展示的发现。