Big Data and Machine Learning Workshop - Day 2 @ UTACM

1
۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬
‫ﺻﺪﯾﻘﯽ‬ ‫ﺍاﻣﯿﺮ‬
‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
@amirsedighi :‫ﺗﻮ4ﺘﺮ‬
:‫ﺍاﯾﻤﯿﻞ‬sedighi@gmail.com
(Unsupervised) ‫‌ﻧﻈﺎﺭرﺕت‬‫ﻥن‬‫ﺑﺪﻭو‬ ‫‌ﻫﺎﯼی‬‫ﺵش‬‫ﺭرﻭو‬ ،‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ - ‫ﺩدﻭوﻡم‬ ‫ﺭرﻭوﺯز‬

2
‫ﺷﺮﻭوﻉع‬ ‫ﺍاﺯز‬ ‫ﻗﺒﻞ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
UTACM-BDML :‫ﻫﺴﺘﯿﻢ‬ ‫ﺍاﺭرﺗﺒﺎﻁط‬ ‫ﺩدﺭر‬ ‫ﺗﻠﮕﺮﺍاﻡم‬ ‫ﻃﺮﯾﻖ‬ ‫ﺍاﺯز‬ ‫ﻫﻢ‬ ‫ﺑﺎ‬ ‫ﺩدﻭوﺭرﻩه‬ ‫ﻃﻮﻝل‬ ‫ﺩدﺭر‬
:‫ﮐﻨﯿﺪ‬ ‫ﺩدﺍاﻧﻠﻮﺩد‬ ‫ﻣﯿﺘﻮﻧﯿﺪ‬ ‫ﻭو‬ ‫ﺍاﺳﻼﯾﺪﺷﺮ‬ ‫ﺭرﻭوﯼی‬ ‫ﻣﯿﺮﻩه‬ ‫ﺟﻠﺴﻪ‬ ‫ﻫﺮ‬ ‫ﺍاﺯز‬ ‫ﺑﻌﺪ‬ ‫‌ﻫﺎ‬‫ﺪ‬‫ﺍاﺳﻼﯾ‬
‫‌ﺷﻮﯾﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﺁآﭼﺎﺭر‬ ‫ﺑﻪ‬ ‫ﺩدﺳﺖ‬ ‫ﮐﻤﯽ‬ ‫ﺟﻠﺴﻪ‬ ‫ﻫﺮ‬ ‫ﺩدﺭر‬ ،‫ﻫﺴﺘﯿﺪ‬ ‫ﺁآﺷﻨﺎ‬ ‫‌ﻧﻮﯾﺴﯽ‬‫ﻪ‬‫ﺑﺮﻧﺎﻣ‬ ‫ﺑﺎ‬ ‫ﮐﻪ‬ ‫ﺍاﯾﻨﻪ‬ ‫ﻣﺎ‬ ‫ﻓﺮﺽض‬
‫‌ﮔﯿﺮﯾﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﯾﺎﺩد‬ ‫ﺳﺮﯾﻊ‬ ‫ﮐﻪ‬ ‫ﺍاﯾﻨﻪ‬ ‫ﻣﺎ‬ ‫ﻓﺮﺽض‬
‫ﮐﻨﯿﺪ‬ ‫ﺳﺎﮐﺖ‬ ‫ﯾﺎ‬ ‫ﺧﺎﻣﻮﺵش‬ ‫ﺭرﻭو‬ ‫‌ﻫﺎ‬‫ﻞ‬‫ﻣﻮﺑﺎﯾ‬ ‫ﻟﻄﻔﺎ‬
‫ﺑﺬﺍاﺭرﯾﻦ‬ ‫ﺧﺎﻟﯽ‬ ‫‌ﺭرﺳﻦ‬‫ﯽ‬‫ﻣ‬ ‫ﺩدﯾﺮﺗﺮ‬ ‫ﺍاﺣﺘﻤﺎﻻ‬ ‫ﮐﻪ‬ ‫ﺍاﻭوﻧﺎﯾﯽ‬ ‫ﺑﺮﺍاﯼی‬ ‫ﺭرﻭو‬ ‫ﮐﻼﺱس‬ ‫ﺩدﺭر‬ ‫ﻧﺰﺩدﯾﮏ‬ ‫‌ﻫﺎﯼی‬‫ﯽ‬‫ﺻﻨﺪﻟ‬ ‫ﻟﻄﻔﺎ‬
‫‌ﮐﻨﯿﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﺍاﯾﻤﯿﻞ‬ ‫ﺑﺮﺍاﻡم‬ ‫ﺑﻌﺪﯼی‬ ‫ﮐﻼﺱس‬ ‫ﺍاﺯز‬ ‫ﻗﺒﻞ‬ ‫ﻭو‬ ‫ﻣﯿﺪﯾﻦ‬ ‫ﺍاﻧﺠﺎﻡم‬ ‫ﮐﻪ‬ ‫ﺩدﺍاﺭرﯾﺪ‬ ‫ﺧﻮﻧﻪ‬ ‫ﺗﻮﯼی‬ ‫ﺑﺮﺍاﯼی‬ ‫ﮐﻮﭼﮏ‬ ‫ﺗﻤﺮﯾﻦ‬ ‫ﯾﮏ‬
‫ﮐﻨﯿﻢ‬ ‫ﻣﻄﺮﺡح‬ ‫ﮔﺮﻭوﻩه‬ ‫ﺩدﺭر‬ ‫ﯾﺎ‬ ‫ﮐﻼﺱس‬ ‫ﺍاﺯز‬ ‫ﺑﻌﺪ‬ ‫ﺭرﻭو‬ ‫ﺑﺤﺮﺍاﻧﯽ‬ ‫ﻏﯿﺮ‬ ‫‌ﻫﺎﯼی‬‫ﺶ‬‫ﭘﺮﺳ‬ .‫ﺩدﺍاﺭرﻩه‬ ‫ﺗﻌﻠﻖ‬ ‫‌ﻫﺎ‬‫ﻪ‬‫ﺑﭽ‬ ‫ﻫﻤﻪ‬ ‫ﺑﻪ‬ ‫ﮐﻼﺱس‬ ‫ﺯزﻣﺎﻥن‬
‫‌ﺍاﻧﺪ‬‫ﻩه‬‫ﺷﺪ‬ ‫ﻣﺸﺨﺺ‬ Q ‫ﺑﺎ‬ ‫ﺍاﺳﻼﯾﺪ‬ ‫ﺩدﺭر‬ ‫ﮐﻪ‬ ‫ﺩدﺍاﺭرﯾﻢ‬ ‫ﮐﻮﭼﮏ‬ ‫‌ﻫﺎﯾﯽ‬‫ﺰ‬4‫ﮐﻮ‬ ‫ﮐﻼﺱس‬ ‫ﻃﻮﻝل‬ ‫ﺩدﺭر‬
‫ﮐﻨﯿﺪ‬ ‫ﺭرﺍاﻫﻨﻤﺎ4ﻢ‬ ‫ﻟﻄﻔﺎ‬ ،‫ﻣﯿﺮﺳﻪ‬ ‫ﺑﮕﻮﺵش‬ ‫ﻧﺎﻣﺎﻧﻮﺱس‬ ‫ﮐﻪ‬ ‫‌ﺑﺮﻡم‬‫ﯽ‬‫ﻣ‬ ‫ﺑﮑﺎﺭر‬ ‫‌ﺍاﯼی‬‫ﻩه‬‫ﻭوﺍاﮊژ‬ ‫ﺍاﮔﺮ‬
slideshare.net/amirsedighi

‫ﺻﺪﯾﻘﯽ‬ ‫ﺍاﻣﯿﺮ‬
:‫ﻣﻮﺳﺲ‬
3
‫ﻣﻌﺮﻓﯽ‬
http://recommender.ir
http://helio.ir
http://commentum.ir
@amirsedighi :‫ﺗﻮ4ﺘﺮ‬
:‫ﺍاﯾﻤﯿﻞ‬sedighi@gmail.com

4
‫ﻣﻘﺪﻣﻪ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬

5
…‫ﺩدﯾﺪﯾﻢ‬ ‫ﺁآﻧﭽﻪ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﺑﺎﻧﻈﺎﺭرﺕت‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬

6
‫‌ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﻣﺎﺷﯿﻦ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
The algorithms used to do
this are very different from
those used for supervised
learning.

7
‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﺩدﺭر‬ ‫‌ﺩدﺍاﺩدﻩه‬‫ﻪ‬‫ﻣﺠﻤﻮﻋ‬ ‫ﺍاﻧﻮﺍاﻉع‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

8
‫ﻫﺴﺘﻨﺪ‬ ‫ﺗﻔﮑﯿﮏ‬ ‫ﻗﺎﺑﻞ‬

9
‫ﻫﺴﺘﻨﺪ‬ ‫ﺁآﺷﻨﺎ‬

10
‫ﺻﺤﯿﺢ‬ ‫ﭘﺎﺳﺦ‬

11
‫ﻭوﺿﻌﯿﺖ‬ ‫‌ﺑﯿﻨﯽ‬‫ﺶ‬‫ﭘﯿ‬ ‫ﺑﺮﺍاﯼی‬ ‫ﺗﺎﺑﻌﯽ‬ ‫ﯾﺎﻓﺘﻦ‬

12

13
‫ﻧﯿﺴﺘﻨﺪ‬ ‫ﺗﻔﮑﯿﮏ‬ ‫ﻗﺎﺑﻞ‬

14
‫ﻫﺴﺘﻨﺪ‬ ‫ﺁآﺷﻨﺎ‬ ‫ﻧﺎ‬

15
‫ﺻﺤﯿﺢ؟؟؟‬ ‫ﭘﺎﺳﺦ‬

16
‫‌ﺩدﺍاﺩدﻩه‬‫ﻪ‬‫ﻣﺠﻤﻮﻋ‬ ‫ﺩدﺭر‬ ‫ﺳﺎﺧﺘﺎﺭر‬ ‫ﻭو‬ ‫ﺭرﺍاﺑﻄﻪ‬ ‫ﯾﺎﻓﺘﻦ‬

17
‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ - ‫ﻣﺎﺷﯿﻦ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

18

19

20

21
‫ﺧﻮﺷﻪ‬

22
Clustering ‫ﯾﺎ‬ ‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﺧﻮﺷ‬
‫ﺧﻮﺷﻪ‬

23
‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﻣﺎﺷﯿﻦ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

24
‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﻣﺎﺷﯿﻦ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫‌ﺳﺎﺧﺘﺎﺭر‬‫ﯽ‬‫ﺑ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﺩدﺭر‬ ‫ﻣﺨﻔﯽ‬ ‫ﺳﺎﺧﺘﺎﺭرﻫﺎﯼی‬ ‫ﯾﺎﻓﺘﻦ‬

25
“Cluster analysis or clustering
is the task of grouping a set of
objects in such a way that
objects in the same group
(called a cluster) are more
similar (in some sense or
another) to each other than to
those in other groups
(clusters).” - wikipedia

26

27
100000 News

28
100000 News
‫ﺧﻮﺩدﮐﺎﺭر‬ ‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﺧﻮﺷ‬

29

30

31

32

33

34
:‫ﺩدﺍاﺭرﻧﺪ‬ ‫‌ﺑﺪﯾﻠﯽ‬‫ﯽ‬‫ﺑ‬ ‫ﻣﺰﺍاﯾﺎﯼی‬ ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﻣﺎﺷﯿﻦ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬

35
(…‫ﺑﺰﺭرﮒگ‬ ‫ﺍاﻋﺪﺍاﺩد‬ ‫ﻗﺎﻧﻮﻥن‬ - ‫ﺗﺼﺎﺩدﻓﯽ‬ - ‫ﻓﺮﺍاﮔﯿﺮﯼی‬ - ‫ﺁآﻣﻮﺯزﺷﯽ‬ ‫‌ﻫﺎﯼی‬‫ﻝل‬‫)ﻣﺜﺎ‬ ‫ﻧﺪﺍاﺭرﻧﺪ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬ ‫ﺑﻪ‬ ‫ﻧﯿﺎﺯز‬

36
!‫ﻧﯿﺴﺘﯿﻢ‬ ‌‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﻣﺠﻤﻮﻋﻪ‬ ‫ﺑﻪ‬ ‫ﻧﺴﺒﺖ‬ ‫ﺍاﻭوﻟﯿﻪ‬ ‫ﺩدﺍاﻧﺶ‬ ‫ﮐﺴﺐ‬ ‫ﻧﯿﺎﺯزﻣﻨﺪ‬

37
!‫ﻧﯿﺴﺘﯿﻢ‬ ‌‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﻣﺠﻤﻮﻋﻪ‬ ‫ﺑﻪ‬ ‫ﻧﺴﺒﺖ‬ ‫ﺍاﻭوﻟﯿﻪ‬ ‫ﺩدﺍاﻧﺶ‬ ‫ﮐﺴﺐ‬ ‫ﻧﯿﺎﺯزﻣﻨﺪ‬
‫ﺭرﻭوﺳﺖ‬ ‫ﭘﯿﺶ‬ ‫ﺁآﻧﭽﻪ‬ ‫ﺍاﺯز‬ ‫ﯾﺎ‬ ،‫ﻧﺪﺍاﺭرﯾﻢ‬ ‫ﮐﺎﻓﯽ‬ ‫ﺁآﺷﻨﺎﯾﯽ‬ ‫ﺩدﺍاﺩدﻩه‬ ‫ﻣﺠﻤﻮﻋﻪ‬ ‫ﺑﺎ‬ ‫ﮐﻪ‬ ‫‌ﻫﺎﯾﯽ‬‫ﻮ‬‫ﺳﻨﺎﺭرﯾ‬ ‫ﺩدﺭر‬
.‫ﻫﺴﺘﻨﺪ‬ ‫ﮐﺎﺭرﺑﺮﺩد‬ ‫ﭘﺮ‬ ‫ﺑﺴﯿﺎﺭر‬ ،‫ﺧﺒﺮﯾﻢ‬ ‫ﺑﯽ‬

38
Unsupervised learning
typically is tasked with
ﬁnding relationships within
data. There are no training
examples used in this
process. Instead, the
system is given a set data
and tasked with ﬁnding
patterns and correlations
therein.

39

40
DNA Microarray Data

41
DNA Microarray Data ‫ﻧﺒﻮﺩدﻥن‬ ‫ﯾﺎ‬ ‫ﺑﻮﺩدﻥن‬ ‫ﺩدﺍاﺭرﺍا‬ ‫ﻣﯿﺰﺍاﻥن‬ ‫ﺍاﺳﺎﺱس‬ ‫ﺑﺮ‬ ‫ﺭرﺍا‬ ‫ﺍاﻓﺮﺍاﺩد‬ ‫ﻣﯿﺨﻮﺍاﻫﯿﻢ‬
‫ﮐﻨﯿﻢ‬ ‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﺩدﺳﺘ‬ ‫ﻣﻌﯿﻦ‬ ‫ﮊژﻥن‬ ‫ﯾﮏ‬

42
DNA Microarray Data
.‫ﺍاﺳﺖ‬ ‫ﺧﺎﺹص‬ ‫ﮊژﻧﯽ‬ ‫ﺍاﺯز‬ ‫ﺍاﻓﺮﺍاﺩد‬ ‫ﺍاﺯز‬ ‫ﯾﮏ‬ ‫ﻫﺮ‬ ‫ﺑﻮﺩدﻥن‬ ‫ﺩدﺍاﺭرﺍا‬ ‫ﻣﯿﺰﺍاﻥن‬ ‫ﺑﯿﺎﻧﮕﺮ‬ ‫ﺭرﻧﮕﻬﺎ‬

43
DNA Microarray Data
.‫ﺍاﺳﺖ‬ ‫ﺧﺎﺹص‬ ‫ﮊژﻧﯽ‬ ‫ﺍاﺯز‬ ‫ﺍاﻓﺮﺍاﺩد‬ ‫ﺍاﺯز‬ ‫ﯾﮏ‬ ‫ﻫﺮ‬ ‫ﺑﻮﺩدﻥن‬ ‫ﺩدﺍاﺭرﺍا‬ ‫ﻣﯿﺰﺍاﻥن‬ ‫ﺑﯿﺎﻧﮕﺮ‬ ‫ﺭرﻧﮕﻬﺎ‬

44
DNA Microarray Data ‫ﺍاﺯز‬ ‫ﺍاﺳﺖ‬ ‫ﺑﺮﺧﻮﺭرﺩدﺍاﺭر‬ ‫‌ﻫﺎﯾﯽ‬‫ﻥن‬‫ﮊژ‬ ‫ﭼﻪ‬ ‫ﺍاﺯز‬ ‫ﻓﺮﺩد‬ ‫ﻫﺮ‬ ‫ﺍاﯾﻨﮑﻪ‬ ‫ﺗﻌ•ﻦ‬ ‫ﺑﺮﺍاﯼی‬
:‫‌ﺷﻮﺩد‬‫ﯽ‬‫ﻣ‬ ‫ﺗﺸﮑﯿﻞ‬ ‫‌ﻫﺎ‬‫ﻥن‬‫ﮊژ‬ ‫ﺍاﺯز‬ ‫‌ﻫﺎﯾﯽ‬‫ﻩه‬‫ﮔﺮﻭو‬ ‫ﻭو‬ ‫‌ﮐﻨﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﺍاﺳﺘﻔﺎﺩدﻩه‬ ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﯾﺎﺩدﯾﮕﺮﯼی‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬

45
DNA Microarray Data ‫ﭼﻨﯿﻦ‬ ‫ﻭوﻟﯽ‬ ،‫ﺩدﺍاﺭرﺩد‬ ‫ﻣﻌﻨﯽ‬ ‫ﭼﻪ‬ ‫ﮔﺮﻭوﻩه‬ ‫ﻫﺮ‬ ‫‌ﺩدﺍاﻧﯿﻢ‬‫ﯽ‬‫ﻧﻤ‬ ‫ﮐﻪ‬ ‫ﺣﺎﻟﯽ‬ ‫ﺩدﺭر‬
:‫ﺩدﺍاﺭرﺩد‬ ‫ﻭوﺟﻮﺩد‬ ‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬

46
DNA Microarray Data ،‫ﺍاﺳﺖ‬ ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺪﻭوﻥن‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬ ‫ﯾﮏ‬ ‫ﺍاﯾﻦ‬
!‫ﺩدﺍاﻧﯿﻢ‬ ‫ﻧﻤﯽ‬ ‫ﭼﯿﺰﯼی‬ ‫ﮔﺮﻭوﻩه‬ ‫ﻫﺮ‬ ‫ﻣﻌﻨﯽ‬ ‫ﺍاﺯز‬ ‫ﺣﺘﯽ‬ ‫ﻭو‬ ‫ﮔﺮﻭوﻩه‬ ‫ﺩدﺭر‬ ‫ﻋﻀﻮﯾﺖ‬ ‫ﻣﻌﻨﯽ‬ ‫ﺍاﺯز‬ ‫ﮐﻪ‬ ‫ﭼﺮﺍا‬

47
DNA Microarray Data
!‫‌ﺍاﯾﻢ‬‫ﻩه‬‫ﻧﺪﺍاﺩد‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﺑﻪ‬ ‫ﺭرﺍا‬ ‫ﺩدﺭرﺳﺘﯽ‬ ‫ﺟﻮﺍاﺏب‬ ‫ﻫﯿﭻ‬

48
‫ﺗﺮﺗﯿﺐ‬ ‫ﺑﺪﯾﻦ‬ .‫ﺑﺎﺷﺪ‬ ‫ﻣﺮﺗﺒﻂ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﺣﺎﻭوﯼی‬ ‫ﮐﺎﻣﭙﯿﻮﺗﺮ‬ ‫ﻫﺮ‬ ‫ﺍاﺳﺖ‬ ‫ﺑﻬﺘﺮ‬ ‫ﺍاﺑﺮﯼی‬ ‫ﭘﺮﺩدﺍاﺯزﺵش‬ ‫ﺩدﺭر‬
.‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﭘﯿﺪﺍا‬ ‫ﺑﻬﺒﻮﺩد‬ ‫ﮐﺎﺭرﺍاﯾﯽ‬

49
…‫ﺩدﺍاﺭرﻧﺪ‬ ‫ﻧﺰﺩدﯾﮑﯽ‬ ‫ﺩدﻭوﺳﺘﯽ‬ ‫ﮐﻪ‬ ‫ﺍاﺟﺘﻤﺎﻋﯽ‬ ‫‌ﻫﺎﯼی‬‫ﻪ‬‫ﺷﺒﮑ‬ ‫ﺩدﺭر‬ ‫ﺍاﻓﺮﺍاﺩد‬ ‫ﺍاﺯز‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﮔﺮﻭو‬ ‫ﺷﻨﺎﺳﺎﯾﯽ‬

50
‫‌ﻣﺸﺘﺮﯾﺎﻥن‬‫ﯼی‬‫ﺑﻨﺪ‬ ‫ﻃﺒﻘﻪ‬ / ‫ﺑﺎﺯزﺍاﺭر‬ ‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﻃﺒﻘ‬

51
‫‌ﻫﺴﺘﯽ‬‫ﻥن‬‫ﺟﻬﺎ‬ ‫ﻭو‬ ‫‌ﻫﺎ‬‫ﻥن‬‫ﮐﻬﮑﺸﺎ‬ ‫ﭘﯿﺪﺍاﯾﺶ‬

52
‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬ ‫ﺑﻪ‬ ‫ﻧﮕﺎﻫﯽ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
:‫ﻣﺴﺌﻠﻪ‬ ‫ﺗﻌﺮﯾﻒ‬
‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺷﮑﻠﯽ‬ ‫ﺑﻪ‬ ‫ﺭرﺍا‬ ‫ﻧﻘﺎﻁط‬ ‫ﺍاﯾﻦ‬ .‫ﺍاﺳﺖ‬ ‫ﺩدﺳﺘﺮﺱس‬ ‫ﺩدﺭر‬ ‌‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﻧﻘﺎﻁط‬ ‫ﻣﺠﻤﻮﻋﻪ‬ ‫ﯾﮏ‬
:‫ﮐﻪ‬ ‫ﮐﻨﯿﺪ‬
.‫ﺑﺎﺷﻨﺪ‬ ‫ﺷﺒﯿﻪ‬ ‫ﯾﮑﺪﯾﮕﺮ‬ ‫ﺑﻪ‬ ‫‌ﮔﯿﺮﻧﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﻗﺮﺍاﺭر‬ ‫ﺧﻮﺷﻪ‬ ‫ﻫﺮ‬ ‫ﺩدﺭر‬ ‫ﮐﻪ‬ ‫ﻧﻘﺎﻃﯽ‬
.‫ﻧﺒﺎﺷﻨﺪ‬ ‫ﺷﺒﯿﻪ‬ ‫ﯾﮑﺪﯾﮕﺮ‬ ‫ﺑﻪ‬ ‫‌ﮔﯿﺮﻧﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﻗﺮﺍاﺭر‬ ‫ﻣﺘﻤﺎﯾﺰ‬ ‫‌ﻫﺎﯼی‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺩدﺭر‬ ‫ﮐﻪ‬ ‫ﻧﻘﺎﻃﯽ‬

53
‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬ ‫ﺑﻪ‬ ‫ﻧﮕﺎﻫﯽ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﯾﮑﺪﯾﮕﺮ‬ ‫ﺑﺎ‬ ‫ﺁآﻧﻬﺎ‬ ‌‫ﺖ‬‫ﺷﺒﺎﻫ‬ ‫ﻭو‬ ‫ﺩدﺍاﺷﺘﻪ‬ ‫ﻗﺮﺍاﺭر‬ ‫‌ﺑﻌﺪﯼی‬‫ﺪ‬‫ﭼﻨ‬ ‫ﺑﺮﺩدﺍاﺭرﯼی‬ ‫ﻓﻀﺎﯼی‬ ‫ﯾﮏ‬ ‫ﺩدﺭر‬ ‫ﻧﻘﺎﻁط‬ ‫ﻣﻌﻤﻮﻻ‬
.‫‌ﺷﻮﺩد‬‫ﯽ‬‫ﻣ‬ ‫ﺍاﻧﺠﺎﻡم‬ (‫ﻓﺎﺻﻠﻪ)ﺍاﻗﻠﯿﺪﺳﯽ‬ ‫ﻣﺤﺎﺳﺒﻪ‬ ‫ﺗﻮﺳﻂ‬

54
‫ﻣﻨﺎﺳﺐ‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﺷﻨﺎﺳﺎﯾﯽ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﺍاﺳﺖ؟‬ ‫ﺗﺮ‬ ‫ﻣﻨﺎﺳﺐ‬ ‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫ﺑﺮﺍاﯼی‬ ‫ﺯزﯾﺮ‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﺍاﺯز‬ ‫ﯾﮏ‬ ‫ﮐﺪﺍاﻡم‬ ،‫ﺍاﻣﻼﮎک‬ ‌‫ﻪ‬‫ﺳﺎﻣﺎﻧ‬ ‫ﯾﮏ‬ ‫ﺩدﺭر‬
x1 : ‫ﻣﺮﺑﻊ‬ ‫ﻣﺘﺮ‬ ‫ﺣﺴﺐ‬ ‫ﺑﺮ‬ ‫ﻣﺴﺎﺣﺖ‬
x2 : ‫ﺧﻮﺍاﺏب‬ ‫ﺍاﻃﺎﻕق‬ ‫ﺗﻌﺪﺍاﺩد‬
x3 : ‫ﺑﻬﺪﺍاﺷﺘﯽ‬ ‫ﺳﺮﻭوﯾﺲ‬ ‫ﺗﻌﺪﺍاﺩد‬
x4 : ‫ﻃﺒﻘﻪ‬
x5 : ‫ﺳﺎﺧﺖ‬ ‫ﺳﺎﻝل‬
x6: ‫ﻣﺤﻠﻪ‬ ‫ﻧﺎﻡم‬
x7: ‫ﭘﺴﺘﯽ‬ ‫ﮐﺪ‬
Q‫ﮔﺮﻓﺖ؟‬ ‫ﮐﺎﺭر‬ ‫ﺑﻪ‬ ‫ﺭرﺍا‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﺣﺪﺍاﮐﺜﺮ‬ ‫‌ﺗﻮﺍاﻥن‬‫ﯽ‬‫ﻣ‬ ‫ﺗﻤﻬﯿﺪﯼی‬ ‫ﭼﻪ‬ ‫ﺑﺎ‬

55
‫‌ﺑﻨﺪﯼی‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺩدﺭر‬ ‌‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﮐﺮﺩدﻥن‬ ‫ﺑﺴﺎﻣﺎﻥن‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﻣﺸﮑﻞ‬ ‫ﺑﺎ‬ ‫ﺭرﺍا‬ ‫ﻣﺎ‬ ”‫ﻣﺤﻠﻪ‬ ‫“ﻧﺎﻡم‬ ‫ﭘﺲ‬ .‫ﺍاﺳﺖ‬ ‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫ﺍاﺳﺎﺱس‬ ‫ﻭو‬ ‫ﭘﺎﯾﻪ‬ ‫ﺍاﻗﻠﯿﺪﺳﯽ‬ ‫ﻣﺤﺎﺳﺒﺎﺕت‬
‫ﻋﺮﺽض‬ ‫ﻭو‬ ‫ﻃﻮﻝل‬ ‫ﺍاﺯز‬ ‫ﺍاﺳﺘﻔﺎﺩدﻩه‬ .‫ﯾﺎﺑﺪ‬ ‫ﺑﻬﺒﻮﺩد‬ ‫ﺍاﺳﺖ‬ ‫ﺑﻬﺘﺮ‬ ”‫“ﮐﺪﭘﺴﺘﯽ‬ ‫ﻫﻤﭽﻨﯿﻦ‬ .‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﻣﻮﺍاﺟﻪ‬
…‫ﺑﺎﺷﺪ‬ ‫ﺧﻮﺑﯽ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬ ‫‌ﺗﻮﺍاﻧﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﺟﻐﺮﺍاﻓﯿﺎﯾﯽ‬
A

56
A
،‫ﺍاﺑﻌﺎﺩد‬ ‫ﺑﺮﺧﯽ‬ ‫ﺑﺮ‬ ‫ﺗﻤﺮﮐﺰ‬ ‫ﺑﺮﺍاﯼی‬
‫ﮐﺮﺩد‬ ‫ﺣﺬﻑف‬ ‫ﺭرﺍا‬ ‫ﺩدﯾﮕﺮ‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﻣﯿﺘﻮﺍاﻥن‬

57
A
،‫ﺍاﺑﻌﺎﺩد‬ ‫ﺑﺮﺧﯽ‬ ‫ﺑﺮ‬ ‫ﺗﻤﺮﮐﺰ‬ ‫ﺑﺮﺍاﯼی‬
‫ﮐﺮﺩد‬ ‫ﺣﺬﻑف‬ ‫ﺭرﺍا‬ ‫ﺩدﯾﮕﺮ‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﻣﯿﺘﻮﺍاﻥن‬
‫ﺑﺮﺍاﯼی‬ ‫ﺍاﺑﻌﺎﺩد‬ ‫ﮐﺎﻫﺶ‬ ‫‌ﻫﺎﯼی‬‫ﮏ‬‫ﺗﮑﻨﯿ‬
‫ﺗﺼﻮﯾﺮ‬ ‫ﻭو‬ ‫ﻣﺤﺎﺳﺒﺎﺕت‬ ‫ﺳﺎﺯزﯼی‬ ‫ﺳﺎﺩدﻩه‬
‫ﻣﻮﺭرﺩد‬ ‫‌ﺗﻮﺍاﻧﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﺑﻬﺘﺮ‬ ‫ﺳﺎﺯزﯼی‬
‫ﮔﯿﺮﻧﺪ‬ ‫ﻗﺮﺍاﺭر‬ ‫ﺍاﺳﺘﻔﺎﺩدﻩه‬

58
SOM creates a discretized low-dimensional representation.

59
‫ﺗﺸﺎﺑﻪ‬ ‫ﻣﺤﺎﺳﺒﻪ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

60
‫ﺗﺸﺎﺑﻪ‬ ‫ﻣﺤﺎﺳﺒﻪ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

61
‫ﻣﯿﻬﻤﺎﻧﯽ‬ ‫ﺩدﺭر‬ ‫ﻫﻤﻬﻤﻪ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
(Coctail Party) ‫‌ﻫﻤﯽ‬‫ﺭر‬‫ﺩدﻭو‬ ‫ﺩدﺭر‬ ‫ﺻﺪﺍا‬ ‫ﺗﻔﮑﯿﮏ‬

62

63

64

65
‫ﺳﺎﺯزﯼی‬ ‫ﭘﯿﺎﺩدﻩه‬ ‫‌ﻫﺎﯼی‬‫ﯼی‬‫ﺩدﺷﻮﺍاﺭر‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

66
‫ﺳﺎﺯزﯼی‬ ‫ﭘﯿﺎﺩدﻩه‬ ‫‌ﻫﺎﯼی‬‫ﯼی‬‫ﺩدﺷﻮﺍاﺭر‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

67
… ‫ﻣﯿﻬﻤﺎﻧﯽ‬ ‫ﺩدﺭر‬ ‫ﻫﻤﻬﻤﻪ‬ ‫ﻣﺴﺌﻠﻪ‬ ‫ﺣﻞ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
http://research.ics.aalto.ﬁ/ica/cocktail/cocktail_en.cgi

68

69

70
‫ﻣﻨﺎﺳﺐ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﮐﻤﮏ‬ ‫ﺑﻪ‬ ‫ﺧﻄﯽ‬ ‫ﯾﮏ‬ ‫‌ﺣﻞ‬‫ﻩه‬‫ﺭرﺍا‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
(Singular Value Decomposition) ‫ﻣﻨﻔﺮﺩد‬ ‫ﻣﻘﺪﺍاﺭر‬ ‫ﺗﺠﺰﯾﻪ‬

71
‫ﻣﻨﺎﺳﺐ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﺑﺎﺷﺪ‬ ‫ﺩدﺷﻮﺍاﺭر‬ ‫ﺑﺴﯿﺎﺭر‬ ‫ﺍاﺳﺖ‬ ‫ﻣﻤﮑﻦ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﯾﮏ‬ ‫ﺗﻮﺳﻌﻪ‬

72
‫ﺑﺎﺷﺪ‬ ‫ﺩدﺷﻮﺍاﺭر‬ ‫ﺑﺴﯿﺎﺭر‬ ‫ﺍاﺳﺖ‬ ‫ﻣﻤﮑﻦ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﯾﮏ‬ ‫ﺗﻮﺳﻌﻪ‬
‫ﺍاﺳﺖ‬ ‫ﻣﺴﺌﻠﻪ‬ ‫‌ﺣﻞ‬‫ﻩه‬‫ﺭرﺍا‬ ‫ﺍاﺯز‬ ‫ﻧﯿﻤﯽ‬ ‫ﻣﻨﺎﺳﺐ‬ ‫ﺗﮑﻨﯿﮏ‬ ‫ﻭو‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬

73
:‫‌ﮐﻨﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﻋﺒﻮﺭر‬ ‫ﺳﻼﻣﺖ‬ ‫ﺑﻪ‬ ‫ﻃﺮﺍاﺣﯽ‬ ‫‌ﻫﺎﯼی‬‫ﺶ‬‫ﭼﺎﻟ‬ ‫ﺑﺴﯿﺎﺭرﯼی‬ ‫ﺍاﺯز‬ ،‫ﻣﻨﺎﺳﺐ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬ ‫ﺑﺎ‬

74
‫ﺷﺪﻩه‬ ‫ﺍاﻧﺠﺎﻡم‬ ‫‌ﻫﺎﯼی‬‫ﺲ‬‫ﯾﻮﺯزﮐﯿ‬ ‫ﻣﻄﺎﻟﻌﻪ‬ ‫ﺍاﻣﮑﺎﻥن‬

75
‫ﭘﺮﻭوﮊژﻩه‬ ‫ﻣﻮﻓﻘﯿﺖ‬ ‫ﺍاﺣﺘﻤﺎﻝل‬ ‫ﻣﯿﺰﺍاﻥن‬ ‫ﺭرﺷﺪ‬

76
‫ﮐﻤﯿﺎﺏب‬ ‫‌ﻫﺎﯼی‬‫ﺺ‬‫ﺗﺨﺼ‬ ‫ﺑﻪ‬ ‫ﻧﯿﺎﺯز‬ ‫ﮐﺎﻫﺶ‬

77
‫ﭘﺮﻭوﮊژﻩه‬ ‫‌ﻫﺎﯼی‬‫ﻪ‬‫ﻫﺰﯾﻨ‬ ‫ﮐﺎﻫﺶ‬

78
‫ﺑﺮﺍاﯼی‬ ‫ﺩدﺭرﺳﺖ‬ ‫ﺭرﻭوﺷﯽ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬
‫ﺩدﺷﻮﺍاﺭر‬ ‫‌ﺍاﯼی‬‫ﻪ‬‫ﻣﺴﺌﻠ‬ ‫ﺣﻞ‬

79
‫‌ﺑﺎﺯز‬‫ﻦ‬‫ﻣﺘ‬ ‫‌ﺣﻞ‬‫ﻩه‬‫ﺭرﺍا‬ ‫ﯾﮏ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

80
‫ﻣﻨﺎﺳﺐ‬ ‫‌ﺑﺎﺯز‬‫ﻦ‬‫ﻣﺘ‬ ‫ﭘﺮﻭوﮊژﻩه‬ ‫ﺍاﻧﺘﺨﺎﺏب‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
:‫ﺧﻮﺏب‬ ‫ﺑﺎﺯز‬ ‫ﻣﺘﻦ‬ ‫ﭘﺮﻭوﮊژﻩه‬ ‫ﺍاﻧﺘﺨﺎﺏب‬
‫ﺑﺮﺍاﯼی‬ ‫ﺩدﺭرﺳﺖ‬ ‫ﺭرﻭوﺷﯽ‬ ‫ﺍاﻧﺘﺨﺎﺏب‬
‫ﺩدﺷﻮﺍاﺭر‬ ‫‌ﺍاﯼی‬‫ﻪ‬‫ﻣﺴﺌﻠ‬ ‫ﺣﻞ‬

81
‫‌ﺑﺎﺯز‬‫ﻦ‬‫ﻣﺘ‬ ‫‌ﺣﻞ‬‫ﻩه‬‫ﺭرﺍا‬ ‫ﯾﮏ‬ ‫ﺑﻪ‬ ‫ﻧﮕﺎﻫﯽ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
2 m

82

83

84

85

86

87

88

89

90
>> help svd

91
K-means ‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫ﺍاﻟﮕﻮﺭرﯾﺘﻢ‬ ‫ﻧﮕﺎﻫﯽ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
.‫ﺍاﺳﺖ‬ K-means ،‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬ ‫ﺗﺮﯾﻦ‬ ‌‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﭘﺮ‬ ‫ﺍاﺯز‬ ‫ﯾﮑﯽ‬
.‫‌ﺍاﻧﺪ‬‫ﻩه‬‫ﺷﺪ‬ ‫ﻣﻨﺸﻌﺐ‬ ‫ﺁآﻥن‬ ‫ﺍاﺯز‬ ‫ﻣﺘﻌﺪﺩدﯼی‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬
.‫ﺍاﺳﺖ‬ ‫ﺳﺮﯾﻊ‬ ‫ﺑﺴﯿﺎﺭر‬
.‫ﺩدﺍاﺭرﺩد‬ ‫ﻭوﺟﻮﺩد‬ ‫ﺑﺎﺯز‬ ‫ﻣﺘﻦ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﭘﺮﻭوﮊژ‬ ‫ﻗﺎﻟﺐ‬ ‫ﺩدﺭر‬ K-means ‫ﮔﻮﻧﺎﮔﻮﻥن‬ ‫‌ﻫﺎﯼی‬‫ﯼی‬‫‌ﺳﺎﺯز‬‫ﻩه‬‫ﭘﯿﺎﺩد‬

92
:‫ﻭوﺭرﻭوﺩدﯼی‬ ‫ﻣﻘﺎﺩدﯾﺮ‬

93
(k) ‫‌ﻫﺎ‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺗﻌﺪﺍاﺩد‬ ‫ﻭو‬ x1, x2, …..xn‫ﻧﻘﺎﻁط‬ ‫ﻣﺠﻤﻮﻋﻪ‬ :‫ﻭوﺭرﻭوﺩدﯼی‬ ‫ﻣﻘﺎﺩدﯾﺮ‬

94
‫ﻓﺮﺿﯽ‬ ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ kc1, c2, …..ck‫‌ﺩدﻫﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﻗﺮﺍاﺭر‬ ‫ﻓﻀﺎ‬ ‫ﺭرﻭوﯼی‬ ‫ﺍاﺗﻔﺎﻗﯽ‬ ‫ﻣﺨﺘﺼﺎﺕت‬ ‫ﺑﺎ‬

95
:‫‌ﮐﻨﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﺗﮑﺮﺍاﺭر‬ ‫ﻫﻤﮕﺮﺍاﯾﯽ‬ ‫ﺣﺼﻮﻝل‬ ‫ﺗﺎ‬
xi ‫ﻫﺮ‬ ‫ﺍاﺯزﺍاﯼی‬ ‫ﺑﻪ‬

96
‫ﻣﯿﺎﺑﯿﻢ‬ ‫ﺭرﺍا‬ cj ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ ‫ﺗﺮﯾﻦ‬ ‫ﻧﺰﺩدﯾﮏ‬

97
‫ﻣﯿﺎﺑﯿﻢ‬ ‫ﺭرﺍا‬ cj ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ ‫ﺗﺮﯾﻦ‬ ‫ﻧﺰﺩدﯾﮏ‬distance(xi, cj)

98
‫‌ﮐﻨﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﻣﻨﺘﺴﺐ‬ j ‫ﺧﻮﺷﻪ‬ ‫ﺑﻪ‬ ‫ﺭرﺍا‬ xi
distance(xi, cj)

99
distance(xi, cj)
:(‫ﻫﺴﺘﻨﺪ‬ ‫ﻋﺪﺩد‬ k ‫)ﮐﻪ‬ j ‫ﺧﻮﺷﻪ‬ ‫ﻫﺮ‬ ‫ﺍاﺯزﺍاﯼی‬ ‫ﺑﻪ‬

100
distance(xi, cj)
‫‌ﮐﻨﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﻣﺤﺎﺳﺒﻪ‬ xi ‫ﻧﻘﺎﻁط‬ ‫ﻣﺘﻮﺳﻂ‬ ‫ﺍاﺳﺎﺱس‬ ‫ﺑﺮ‬ ‫ﺟﺪﯾﺪﯼی‬ ‫ﻣﺤﻞ‬ cj ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ ‫ﺑﺮﺍاﯼی‬

101
distance(xi, cj)
Q‫ﭼﺮﺍا؟‬ ‫ﻭو‬ ‫ﮐﻨﯿﻢ‬ ‫ﺍاﺳﺘﻔﺎﺩدﻩه‬ ‫‌ﺗﻮﺍاﻧﯿﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﺭرﺍا‬ ‫ﺍاﺑﻌﺎﺩدﯼی‬ ‫ﭼﻪ‬

102
distance(xi, cj)
.‫‌ﺷﻮﯾﻢ‬‫ﯽ‬‫ﻣ‬ ‫ﻣﺘﻮﻗﻒ‬ ،‫ﻧﺸﺪ‬ ‫ﺣﺎﺻﻞ‬ ‫ﺗﻐ•ﺮﯼی‬ ‫ﺍاﮔﺮ‬

103
distance(xi, cj)
QO?

104
distance(xi, cj)
O(‫‌ﻫﺎ*ﻧﻘﺎﻁط*ﺍاﺑﻌﺎﺩد*ﺗﮑﺮﺍاﺭر‬‫ﻪ‬‫)ﺧﻮﺷ‬

105

106

107

108

109

110

111

112

113

114

115
K-means ‫ﺿﻌﻒ‬ ‫ﻭو‬ ‫ﻗﻮﺕت‬ ‫ﻧﻘﺎﻁط‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
:‫ﮐﻨﯿﺪ‬ ‫ﺫذﮐﺮ‬ ‫ﺭرﺍا‬ K-means ‫ﺿﻌﻒ‬ ‫ﻧﻘﺎﻁط‬ Q

116
:K-means ‫ﺿﻌﻒ‬ ‫ﻧﻘﺎﻁط‬ A

117
‫ﺍاﺧﺘﻼﻝل‬ ‫ﻭو‬ (Outlier) ‫ﻣﺤﺪﻭوﺩدﻩه‬ ‫ﺍاﺯز‬ ‫ﺧﺎﺭرﺝج‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﺑﻪ‬ ‫ﺑﺎﻻ‬ ‫ﺑﺴﯿﺎﺭر‬ ‫ﺣﺴﺎﺳﯿﺖ‬

118
‫ﺍاﺧﺘﻼﻝل‬ ‫ﻭو‬ (Outlier) ‫ﻣﺤﺪﻭوﺩدﻩه‬ ‫ﺍاﺯز‬ ‫ﺧﺎﺭرﺝج‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﺑﻪ‬ ‫ﺑﺎﻻ‬ ‫ﺑﺴﯿﺎﺭر‬ ‫ﺣﺴﺎﺳﯿﺖ‬
‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻧﻤ‬ ‫ﺗﻮﻟﯿﺪ‬ ‫ﺧﻮﺑﯽ‬ ‫‌ﻫﺎﯼی‬‫ﺏب‬‫ﺟﻮﺍا‬ ‫ﻧﯿﺴﺘﻨﺪ‬ (‌‫ﻩه‬‫)ﺩدﺍاﯾﺮ‬ ‫ﮐﺮﻭوﯼی‬ ‫ﮐﻪ‬ ‫‌ﻫﺎﯾﯽ‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺑﺮﺍاﯼی‬

119
:‫ﮐﻨﯿﺪ‬ ‫ﺫذﮐﺮ‬ ‫ﺭرﺍا‬ K-means ‫ﻗﻮﺕت‬ ‫ﻧﻘﺎﻁط‬ Q

120
:K-means ‫ﻗﻮﺕت‬ ‫ﻧﻘﺎﻁط‬ A

121
A
‫ﺍاﺳﺖ‬ ‫ﺁآﺳﺎﻥن‬ ‫ﺁآﻥن‬ ‫ﻓﻬﻤﯿﺪﻥن‬
:K-means ‫ﻗﻮﺕت‬ ‫ﻧﻘﺎﻁط‬

122
A
‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﮐﺎﺭر‬ ‫ﺧﻮﺑﯽ‬ ‫ﺑﻪ‬ ‫ﺑﺰﺭرﮒگ‬ ‫ﺑﺴﯿﺎﺭر‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫ﻣﺠﻤﻮﻋﻪ‬ ‫ﺭرﻭوﯼی‬

123
A
(۱۰۰x) ‫ﺍاﺳﺖ‬ ‫ﮐﻢ‬ ‫ﻣﺤﺴﻮﺳﯽ‬ ‫ﻃﻮﺭر‬ ‫ﺑﻪ‬ ‫ﺩدﻗﯿﻘﺘﺮ‬ ‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺍاﻟﮕﻮﺭرﯾﺘ‬ ‫ﺑﻪ‬ ‫ﻧﺴﺒﺖ‬ ‫ﺁآﻥن‬ ‫ﺍاﺟﺮﺍاﯼی‬ ‫ﻫﺰﯾﻨﻪ‬

124
A
‫ﺩدﺍاﺭرﺩد‬ ‫ﻧﯿﺎﺯز‬ (k) ‫‌ﻫﺎ‬‫ﻪ‬‫ﺧﻮﺷ‬ ‫ﺗﻌﺪﺍاﺩد‬ ‫ﺑﻪ‬ ‫ﻓﻘﻂ‬ ‫ﺍاﺟﺮﺍا‬ ‫ﺑﺮﺍاﯼی‬

125
A
‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﺳﺎﺩدﻩه‬ ‫ﺭرﺍا‬ ‫ﻧﺘﯿﺠﻪ‬ ،‫‌ﺷﻮﺩد‬‫ﯽ‬‫ﻣ‬ ‫ﺍاﺭرﺍاﺋﻪ‬ ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ ‫ﻣﻌﺮﻓﯽ‬ ‫ﺗﻮﺳﻂ‬ ‫ﮐﻪ‬ ‫ﺑﺎﻻﯾﯽ‬ ‫ﺗﺠﺮﯾﺪ‬

126
A
‫‌ﮐﻨﺪ‬‫ﯽ‬‫ﻣ‬ ‫ﺳﺎﺩدﻩه‬ ‫ﺭرﺍا‬ ‫ﻧﺘﯿﺠﻪ‬ ،‫‌ﺷﻮﺩد‬‫ﯽ‬‫ﻣ‬ ‫ﺍاﺭرﺍاﺋﻪ‬ ‫ﺧﻮﺷﻪ‬ ‫ﻣﺮﮐﺰ‬ ‫ﻣﻌﺮﻓﯽ‬ ‫ﺗﻮﺳﻂ‬ ‫ﮐﻪ‬ ‫ﺑﺎﻻﯾﯽ‬ ‫ﺗﺠﺮﯾﺪ‬
‫ﺍاﺳﺖ‬ ‫ﺳﺎﺩدﻩه‬ ‫ﺑﺴﯿﺎﺭر‬ ‫ﺁآﻥن‬ ‫ﺳﺎﺯزﯼی‬ ‫ﭘﯿﺎﺩدﻩه‬

127
‫ﺳﺎﺯزﯼی‬ ‫ﭘﯿﺎﺩدﻩه‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﺑﻨﻮﯾﺴﯿﻢ‬ ‫ﮐﺪ‬

128
‫ﺳﺎﺯزﯼی‬ ‫ﭘﯿﺎﺩدﻩه‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﺩدﻧﺒﺎﻝل‬ ‫ﺑﻪ‬ ‫ﻭوﺭرﺯزﺷﯽ‬ ‫ﺍاﺧﺒﺎﺭر‬ ‫ﺑﻨﮕﺎﻩه‬ ‫ﯾﮏ‬
‫‌ﻫﺎﯼی‬‫ﻢ‬‫ﺗﯿ‬ ‫ﺑﯿﻦ‬ ‫ﺷﻬﺮﺁآﻭوﺭرﺩد‬ ‫ﯾﮏ‬ ‫ﺑﺮﮔﺰﺍاﺭرﯼی‬
‫ﺑﻨﺪﯼی‬ ‫ﺧﻮﺷﻪ‬ ‫ﻧﯿﺎﺯزﻣﻨﺪ‬ ،‫ﻗﺮﻣﺰ‬ ‫ﻭو‬ ‫ﺁآﺑﯽ‬
.‫ﺍاﺳﺖ‬ ‫ﻣﺮﺗﺒﻂ‬ ‫ﺍاﺧﺒﺎﺭر‬

129
‫ﻣﻠﺰﻭوﻣﺎﺕت‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
Prerequisites:
• Windows, Linux or Mac
• Java 1.7
• Apache Maven 3

130
CM - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
Create the Maven project:
mvn archetype:generate
-DarchetypeGroupId=org.apache.maven.archetypes
-DgroupId=com.technobium
-DartifactId=mahout-clustering
-DinteractiveMode=false

131
‫ﻧﺎﻡم‬ ‫ﺮ‬e‫ﺗﻐ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
Rename the default created App class to ClusteringDemo
mv mahout-clustering/src/main/java/com/technobium/App.java
mahout-clustering/src/main/java/com/technobium/ClusteringDemo

132
‫‌ﻫﺎ‬‫ﯽ‬‫ﻭوﺍاﺑﺴﺘﮕ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
Add the Mahout and SLF4J libraries to this project:
cd mahout-clustering
nano pom.xml
<dependency>
<groupId>org.apache.mahout</groupId>
<artifactId>mahout-core</artifactId>
<version>0.9</version>
</dependency>
<dependency>
<groupId>org.slf4j</groupId>
<artifactId>slf4j-simple</artifactId>
<version>1.7.7</version>
</dependency>

133
‫ﮐﺪ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

134
‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
‫ﮐﺪ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬

135

136

137
‫ﺍاﺟﺮﺍا‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
mvn compile
mvn exec:java -Dexec.mainClass="com.technobium.ClusteringDemo"
Run the ClusteringDemo class by using the following commands:

138
‫ﻧﺘﯿﺠﻪ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬

139
‫‌ﻫﺎﺏب‬‫ﺖ‬‫ﮔﯿ‬ - ‫ﻧﻈﺎﺭرﺕت‬ ‫ﺑﺎ‬ ‫ﯾﺎﺩدﮔﯿﺮﯼی‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
https://github.com/technobium/mahout-clustering

140
TFIDF – term frequency inverse document frequency is an important weighting
scheme which can be used in ﬁelds like machine learning, natural language
processing, search engines and text mining. The metric is used to measure the
relative importance of a word for a collection of documents. If a term or word
occurs frequently in a document and not so frequently in the entire set of
documents, it is more relevant to a search than a word that appears frequently
across all the documents. By calculating TFIDF for all terms which appear in a
set of document we can ﬁlter away the less relevant words. As an example, a
word which appears only twice in a single document is more relevant to
someone searching that document, compare to words which appear many
times in all the documents like: the, is, at, and, or, on, etc. Using TFIDF the later
words can be ignored and the relevant ones are retained.

141
The canopy clustering algorithm is an unsupervised pre-clustering
algorithm introduced by Andrew McCallum, Kamal Nigam and Lyle
Ungar in 2000. It is often used as preprocessing step for the K-means
algorithm or the Hierarchical clustering algorithm. It is intended to speed
up clustering operations on large data sets, where using another
algorithm directly may be impractical due to the size of the data set.

142
‫ﻣﺮﺟﻊ‬‫ﺍاﻥن‬‫ﺮ‬‫ﺗﻬ‬ ‫ﺩدﺍاﻧﺸﮕﺎﻩه‬ ACM - ۱۳۹۵ ‫ﺗﺎﺑﺴﺘﺎﻥن‬ - ‫‌ﻣﺎﺷﯿﻦ‬‫ﯼی‬‫ﯾﺎﺩدﮔﯿﺮ‬ ‫ﻭو‬ ‫ﺑﺰﺭرﮒگ‬ ‫‌ﻫﺎﯼی‬‫ﻩه‬‫ﺩدﺍاﺩد‬ ‫‌ﻫﺎﯼی‬‫ﺩد‬‫ﮐﺎﺭرﺑﺮ‬ ‫ﺑﺮ‬ ‫ﮔﺬﺭرﯼی‬
http://www.datavizualization.com/blog/an-introduction-to-machine-learning-theory-and-its-applications-a
https://www.coursera.org/learn/machine-learning
https://en.wikipedia.org/wiki/Cluster_analysis
http://stackoverﬂow.com/questions/20414667/cocktail-party-algorithm-svd-implementation-in-one-line-of-code
https://www.gnu.org/software/octave/
http://technobium.com/introduction-to-clustering-using-apache-mahout/
https://github.com/technobium/mahout-clustering
http://homepages.inf.ed.ac.uk/vlavrenk/
http://stats.stackexchange.com/questions/58855/why-do-we-use-k-means-instead-of-other-algorithms
http://mahout.apache.org/users/clustering/canopy-clustering.html
http://mahout.apache.org/users/clustering/fuzzy-k-means.html
https://en.wikipedia.org/wiki/Canopy_clustering_algorithm

Big Data and Machine Learning Workshop - Day 2 @ UTACM

More Related Content

Viewers also liked (16)

More from Amir Sedighi (10)

Big Data and Machine Learning Workshop - Day 2 @ UTACM