From f3fda1b753616ab74fb177cb316dd46f667a4bc4 Mon Sep 17 00:00:00 2001 From: lintong <40327910+kjlintong@users.noreply.github.com> Date: Wed, 10 Jul 2019 20:26:15 +0800 Subject: [PATCH 1/4] =?UTF-8?q?Create=2011=E3=80=81=E6=80=BB=E7=BB=93?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- "doc/11\343\200\201\346\200\273\347\273\223" | 1 + 1 file changed, 1 insertion(+) create mode 100644 "doc/11\343\200\201\346\200\273\347\273\223" diff --git "a/doc/11\343\200\201\346\200\273\347\273\223" "b/doc/11\343\200\201\346\200\273\347\273\223" new file mode 100644 index 0000000..8b13789 --- /dev/null +++ "b/doc/11\343\200\201\346\200\273\347\273\223" @@ -0,0 +1 @@ + From 16f8795163b8250342c08613fa58f9990437560a Mon Sep 17 00:00:00 2001 From: lintong <40327910+kjlintong@users.noreply.github.com> Date: Wed, 10 Jul 2019 20:26:38 +0800 Subject: [PATCH 2/4] =?UTF-8?q?Rename=2011=E3=80=81=E6=80=BB=E7=BB=93=20to?= =?UTF-8?q?=2011=E3=80=81=E6=80=BB=E7=BB=93.md?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- .../11\343\200\201\346\200\273\347\273\223.md" | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename "doc/11\343\200\201\346\200\273\347\273\223" => "doc/11\343\200\201\346\200\273\347\273\223.md" (100%) diff --git "a/doc/11\343\200\201\346\200\273\347\273\223" "b/doc/11\343\200\201\346\200\273\347\273\223.md" similarity index 100% rename from "doc/11\343\200\201\346\200\273\347\273\223" rename to "doc/11\343\200\201\346\200\273\347\273\223.md" From 7c237e7c7b84c1320fc14ce0937a21942be70b7b Mon Sep 17 00:00:00 2001 From: lintong <40327910+kjlintong@users.noreply.github.com> Date: Thu, 25 Jul 2019 20:38:08 +0800 Subject: [PATCH 3/4] =?UTF-8?q?Update=2011=E3=80=81=E6=80=BB=E7=BB=93.md?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- "doc/11\343\200\201\346\200\273\347\273\223.md" | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git "a/doc/11\343\200\201\346\200\273\347\273\223.md" "b/doc/11\343\200\201\346\200\273\347\273\223.md" index 8b13789..1e50126 100644 --- "a/doc/11\343\200\201\346\200\273\347\273\223.md" +++ "b/doc/11\343\200\201\346\200\273\347\273\223.md" @@ -1 +1,11 @@ +# 其它形式的学习 +## 度量学习 + +我曾提到过衡量两个特征向量之间相似性(或差异)最常用的度量是欧氏距离和余弦相似性。这种度量选择似乎是符合逻辑,但实际上是随意的,就像在线性回归中选择平方误差一样。事实上,一个指标可以比另一个指标性能更好,取决于特定的数据集。也就是说,没有一种指标是永远完美的。 + +您可以创建更适合您的数据集的指标。然后可以将您的度量标准集成到任何需要度量标准的学习算法中,例如 k-means 或 kNN。您如何在不尝试所有可能性的情况下知道哪个方程式是一个好的度量标准?您可以用数据训练您的度量标准。 + +记住两个向量 x 和 x`间的欧氏距离: + + From 8f66146d23328ee4e9b9a0f2b5180940245bdcb4 Mon Sep 17 00:00:00 2001 From: kjlintong <979866078@qq.com> Date: Fri, 26 Jul 2019 15:34:11 +0800 Subject: [PATCH 4/4] =?UTF-8?q?Update=2011=E3=80=81=E6=80=BB=E7=BB=93.md?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- "doc/11\343\200\201\346\200\273\347\273\223.md" | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git "a/doc/11\343\200\201\346\200\273\347\273\223.md" "b/doc/11\343\200\201\346\200\273\347\273\223.md" index 1e50126..3a0c450 100644 --- "a/doc/11\343\200\201\346\200\273\347\273\223.md" +++ "b/doc/11\343\200\201\346\200\273\347\273\223.md" @@ -8,4 +8,5 @@ 记住两个向量 x 和 x`间的欧氏距离: - + +$$ d_{\mathbf{A}}\left(\mathbf{x},\mathbf{x}^{\prime}\right)=\left\|\mathbf{x}-\mathbf{x}^{\prime}\right\| \mathbf{A} \stackrel{\mathrm{def}}{=} \sqrt{\left(\mathbf{x}-\mathbf{x}^{\prime}\right)^{\top} \mathbf{A}\left(\mathbf{x}-\mathbf{x}^{\prime}\right)} $$