際際滷
Submit Search
ESM SVM & Unsupervised Learning
0 likes
231 views
Mario Cho
Follow
ESM( 襾語 ) 8譯殊姶 SVM, Unsupervised Learning
Read less
Read more
1 of 49
Download now
Download to read offline
More Related Content
ESM SVM & Unsupervised Learning
1.
Mario Cho(譟 襷) Hephaex@gmail.com
3.
Source: http://kosis.kr/statHtml/statHtml.do?orgId=101&tblId=DT_1HDCA10
4.
蟲 蟲覩
谿譟壱 螳蟲 : 3 6豌襷 蟲螳 豌 豐 一 螳蟲襦 蟆 蠍一 覯瑚骸 覿(覿 覯語!) 觜殊 蟲覩朱れ 螳蟲 豐 一 . 糾豌 5覿覲 覦 覿豈 覲 螳蟲 譟一 蠏 28,065 襷 ( 34,246襷 覿豈 6,181襷) 譴螳螳 16,194 襷 ( 20,150襷 覿豈 4,470襷) 1覿 蠏磯轟 54.2 2,3,4覿螳 51 5覿 蠏磯轟 54.9 1% 139,200( 蠍 覿) 2% 117,545( 蠍 覿) 4.2% 100,000 5% 89,900 10% 61,800 15% 50,000 20% 42,000 30% 30,700 40% 22,700 50% 16,200 60% 11,500 70% 7,500 80% 3,700 90% 1,030 Y= 27638ln(x) + 124674
8.
y(x)=+1 y(x)=0 y(x)=-1 margin = | | ||Max襦 螻
矩! 覿襯襯 螳襦讌襯企 蟆所譟郁唄 y(x)=0螻 y(x) = +1, 轟 y(x) = -1 蟆所 譟郁唄螳 螳蟆(margin) 豕 蟆 覈. y(x) = 0 蟆所 譟郁唄 一危 x 蟾讌 蠍壱 蟇磯Μ(, 轟 觜) | | || 襦 . margin 豕 y(x)=0 蟆所 譟郁唄 谿城 蟆 SVM 覈企.
9.
y(x)=0 q x1
10.
y(x)=0 q x1
11.
y(x)=0 q x1 QTX > +1
12.
y(x)=0 q x1 x2
13.
y(x)=0 q x1 x2
14.
y(x)=0 q x1 x2
15.
y(x)=0 q x1 x2
16.
螳 x
g(z) = QTx + b 襦 螻. g(z) 襯 螳 x 螳螳 伎 螻壱 g(z) > +1 : g(z) < -1 : 蠏れ y(x)=0 q x1 x2 QTX2 < -1
17.
螳 x
g(z) = QTx + b 襦 . g(z) 襯 螳 x 螳螳 kernel 牛 螻壱 g(z) > +1 : g(z) < -1 : 蠏れ 蟲り . y(x)=0 L1 L2
18.
螳 x
g(z) = QTx + b 襦 . g(z) 襯 螳 x 螳螳 kernel 牛 螻壱 g(z) > +1 : g(z) < -1 : 蠏れ 蟲り . y(x)=0 L1 L2
19.
y(x)=0 L1 L2
20.
L =
5朱 螳一 覿
21.
L =
5朱 螳一 覿 蠍一 襦 覦蠖 蠏碁?
23.
y(x)=0 L1 L2
24.
y(x)=0 L1 L2
25.
y(x)=0 L1 L2
26.
y(x)=0 L1 L2
27.
y(x)=0 L1 L2
28.
y(x)=0 L1 L2 y(x)=-1 y(x)=+1
29.
y(x)=0 L1 L2 y(x)=-1 y(x)=+1
30.
y(x)=0 L1 L2 y(x)=-1 y(x)=+1
40.
Cluster Assignment move centroid
Download