HackerNews中文版HackerNews CN
GitHubTwitter

矩阵变换器：无需训练的结构模式发现

1 分•作者: AyodeleFikayomi•7 个月前•原帖
扫描原始矩阵以寻找符号结构（例如，秩、对称性、熵）——无需标签，无需训练。 适用于多个领域（生物学、文本、图形）。 可以在以下链接中尝试： https://mybinder.org/v2/gh/fikayoAy/MatrixTransformer/HEAD?filepath=run_demo.ipynb https://colab.research.google.com/github/fikayoAy/MatrixTransformer/blob/main/run_demo.ipynb GitHub链接： https://github.com/fikayoAy/MatrixTransformer 非常适合数据理解、异常检测或算法选择。
查看原文
Scans raw matrices to find symbolic structure (e.g., rank, symmetry, entropy) — no labels, no training required. Works across domains (biology, text, graphs). Try it in: https:&#x2F;&#x2F;mybinder.org&#x2F;v2&#x2F;gh&#x2F;fikayoAy&#x2F;MatrixTransformer&#x2F;HEAD?filepath=run_demo.ipynb https:&#x2F;&#x2F;colab.research.google.com&#x2F;github&#x2F;fikayoAy&#x2F;MatrixTransformer&#x2F;blob&#x2F;main&#x2F;run_demo.ipynb<p>Github Link: https:&#x2F;&#x2F;github.com&#x2F;fikayoAy&#x2F;MatrixTransformer Great for data understanding, anomaly detection, or algorithm selection.