Upload File

introduction to machine learning - college of · pdf fileintroduction)to) machine)learning...

  • Home
  • Documents
  • Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins [email protected]
20
INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2010 Edited and expanded for CS 4641 by Chris Simpkins [email protected] h1p://www.cmpe.boun.edu.tr/~ethem/i2ml2e Lecture Slides for 1

Upload: lynhi

Post on 23-Mar-2018

241 views

Category:

Documents


4 download

Report

TRANSCRIPT

Page 1: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

INTRODUCTION  TO  Machine  Learning

ETHEM  ALPAYDIN©  The  MIT  Press,  2010

Edited  and  expanded  for  CS  4641  by  Chris  Simpkins

[email protected]://www.cmpe.boun.edu.tr/~ethem/i2ml2e

Lecture  Slides  for

1

Page 2: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

CHAPTER  1:  IntroducOon

2

Page 3: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Why  “Learn”  ? Machine  learning  is  programming  computers  to  opOmize  a  performance  criterion  using  example  data  or  past  experience.

There  is  no  need  to  “learn”  to  calculate  payroll Learning  is  used  when:

Human  experOse  does  not  exist  (navigaOng  on  Mars), Humans  are  unable  to  explain  their  experOse  (speech  recogniOon)

SoluOon  changes  in  Ome  (rouOng  on  a  computer  network) SoluOon  needs  to  be  adapted  to  parOcular  cases  (user  biometrics)

3

3

Page 4: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

What  We  Talk  About  When  We    Talk  About“Learning” Learning  general  models  from  a  data  of  parOcular  examples  

Data  is  cheap  and  abundant  (data  warehouses,  data  marts);  knowledge  is  expensive  and  scarce.  

Example  in  retail:  Customer  transacOons  to  consumer  behavior:     People  who  bought  “Blink”  also  bought  “Outliers”    (www.amazon.com)

Build  a  model  that  is  a  good  and  useful  approximaGon  to  the  data.    

4

4

Page 5: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Data  Mining Retail:  Market  basket  analysis,  Customer  relaOonship  management  (CRM)

Finance:  Credit  scoring,  fraud  detecOon Manufacturing:  Control,  roboOcs,  troubleshooOng Medicine:  Medical  diagnosis TelecommunicaOons:  Spam  filters,  intrusion  detecOon BioinformaOcs:  MoOfs,  alignment Web  mining:  Search  engines ...

5

5

Page 6: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

What  is  Machine  Learning? OpOmize  a  performance  criterion  using  example  data  or  past  experience.

Role  of  StaOsOcs:  Inference  from  a  sample Role  of  Computer  science:  Efficient  algorithms  to

Solve  the  opOmizaOon  problem RepresenOng  and  evaluaOng  the  model  for  inference

6

6

Page 7: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Types  of  Machine  Learning AssociaOon Supervised  Learning

ClassificaOon Regression

Unsupervised  Learning Reinforcement  Learning

7

7

Page 8: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Learning  AssociaOons Basket  analysis:     P  (Y  |  X  )  probability  that  somebody  who  buys  X  also  buys  Y  where  X  and  Y  are  products/services.

    Example:  P  (  chips  |  beer  )  =  0.7

8

8

Page 9: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

ClassificaOon Example:  Credit  scoring

DifferenOaOng  between  low-­‐risk  and  high-­‐risk  customers  from  their  income  and  savings

Discriminant:  IF  income  >  θ1  AND  savings  >  θ2           THEN  low-­‐risk  ELSE  high-­‐risk

9

9

Page 10: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

ClassificaOon:  ApplicaOons Aka  Pamern  recogniOon Face  recogniOon:  Pose,  lighOng,  occlusion  (glasses,  beard),  make-­‐up,  hair  style  

Character  recogniOon:  Different  handwriOng  styles. Speech  recogniOon:  Temporal  dependency.   Medical  diagnosis:  From  symptoms  to  illnesses Biometrics:  RecogniOon/authenOcaOon  using  physical  and/or  behavioral  characterisOcs:  Face,  iris,  signature,  etc

...

10

10

Page 11: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Face  RecogniOonTraining examples of a person

Test images

ORL dataset,AT&T Laboratories, Cambridge UK

11

11

Page 12: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Regression Example:  Price  of  a  used  car

x  :  car  amributes   y  :  price     y  =  g  (x  |  θ )   g  (  )  model, θ parameters

y  =  wx+w0

12

12

Page 13: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Regression  ApplicaOons NavigaOng  a  car:  Angle  of  the  steering KinemaOcs  of  a  robot  arm

α1=  g1(x,y)α2=  g2(x,y)

α1

α2

(x,y)

n Response  surface  design

13

13

Page 14: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Supervised  Learning:  Uses PredicOon  of  future  cases:  Use  the  rule  to  predict  the  output  for  future  inputs

Knowledge  extracOon:  The  rule  is  easy  to  understand Compression:  The  rule  is  simpler  than  the  data  it  explains Outlier  detecOon:  ExcepOons  that  are  not  covered  by  the  rule,  e.g.,  fraud

14

14

Page 15: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Unsupervised  Learning Learning  “what  normally  happens” No  output Clustering:  Grouping  similar  instances Example  applicaOons

Customer  segmentaOon  in  CRM Image  compression:  Color  quanOzaOon BioinformaOcs:  Learning  moOfs

15

15

Page 16: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Reinforcement  Learning Learning  a  policy:  A  sequence  of  outputs No  supervised  output  but  delayed  reward Credit  assignment  problem Game  playing Robot  in  a  maze MulOple  agents,  parOal  observability,  ...

16

16

Page 17: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Resources:  Datasets UCI  Repository:  hmp://www.ics.uci.edu/~mlearn/MLRepository.html

UCI  KDD  Archive:  hmp://kdd.ics.uci.edu/summary.data.applicaOon.html

Statlib:  hmp://lib.stat.cmu.edu/

Delve:  hmp://www.cs.utoronto.ca/~delve/

17

17

http://www.ics.uci.edu/~mlearn/MLRepository.html
http://www.ics.uci.edu/~mlearn/MLRepository.html
http://kdd.ics.uci.edu/summary.data.application.html
http://kdd.ics.uci.edu/summary.data.application.html
http://kdd.ics.uci.edu/summary.data.application.html
http://kdd.ics.uci.edu/summary.data.application.html
http://lib.stat.cmu.edu/
http://lib.stat.cmu.edu/
http://www.cs.utoronto.ca/~delve/
http://www.cs.utoronto.ca/~delve/
Page 18: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Resources:  Journals Journal  of  Machine  Learning  Research  www.jmlr.org Machine  Learning   Neural  ComputaOon Neural  Networks IEEE  TransacOons  on  Neural  Networks IEEE  TransacOons  on  Pamern  Analysis  and  Machine  Intelligence

Annals  of  StaOsOcs Journal  of  the  American  StaOsOcal  AssociaOon ...

18

18

http://www.jmlr.org/
http://www.jmlr.org/
Page 19: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Resources:  Conferences InternaOonal  Conference  on  Machine  Learning  (ICML)   European  Conference  on  Machine  Learning  (ECML) Neural  InformaOon  Processing  Systems  (NIPS) Uncertainty  in  ArOficial  Intelligence  (UAI) ComputaOonal  Learning  Theory  (COLT) InternaOonal  Conference  on  ArOficial  Neural  Networks  (ICANN)   InternaOonal  Conference  on  AI  &  StaOsOcs  (AISTATS) InternaOonal  Conference  on  Pamern  RecogniOon  (ICPR) ...

19

19

Page 20: Introduction to Machine Learning - College of · PDF fileINTRODUCTION)TO) Machine)Learning ETHEMALPAYDIN ©)The)MIT)Press,)2010 Edited)and)expanded)for)CS)4641)by)Chris)Simpkins alpaydin@boun.edu.tr

Lecture  Notes  for  E  Alpaydın  2010  IntroducOon  to  Machine  Learning  2e  ©  The  MIT  Press  (V1.0)

Resources:  Georgia  Techh"p://ml.cc.gatech.edu/

Charles  Isbell  (my  advisor) Reinforcement  Learning

Andrea  Thomaz Social  Learning

Alex  Gray StaFsFcal  Learning  Theory,  ML  SoIware

Irfan  Essa AcFvity  RecogniFon

James  Rehg Graphical  Models

Maria  Balcan Machine  Learning  Theory

Frank  Dellaert Computer  Vision

20

20

http://ml.cc.gatech.edu/
http://ml.cc.gatech.edu/

SFH 4641 - dammedia.osram.info · SFH 4641 11 Version 1.5 | 2018-05-30 Barcode-Product-Label (BPL) Moisture-sensitive product is packed in a dry bag containing desiccant and a humidity

4641 4645.output

ETHEM ALPAYDIN © The MIT Press, 2010 [email protected] ethem/i2ml2e Lecture Slides for 1 Lecture Notes for E Alpaydın 2010

INTRODUCTION TO Machine Learning - CmpE WEBethem/i2ml/slides/v1... · INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 [email protected] ethem/i2ml Lecture

1 Classification & Clustering 魏志達 Jyh-Da Wei -- Parametric and Nonparametric Methods Introduction to Machine Learning (Chap 4,5,7,8), E. Alpaydin

INTRODUCTION TO MACHINE LEARNING 3RD EDITION ETHEM ALPAYDIN Modified by Prof. Carolina Ruiz © The MIT Press, 2014 for CS539 Machine Learning at WPI [email protected]

INTRODUCTION TO Machine Learningethem/i2ml/slides/v1... · INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 [email protected] ethem/i2ml Lecture Slides for

FOR SALE BIG BOY RESTAURANT 4641 SOUTH RED ARROW …€¦ · FOR SALE. BIG BOY RESTAURANT. 4641 SOUTH RED ARROW HWY STEVENSVILLE, MI 49127. Committed to Brokerage Excellence for over

Howlin brendan wexford news 4641

Combining multiple learners Usman Roshan. Decision tree From Alpaydin, 2010

Introduction to Machine Learning - Ethem Alpaydin

Lecture Notes for E Alpaydın 2010 Introduction to Machine Learning 2e © The MIT Press (V1.0) ETHEM ALPAYDIN © The MIT Press, 2010 [email protected]

2015-2016 Bill 4641: Department of Behavioral and Public ...scstatehouse.gov/sess121_2015-2016/bills/4641.docx  · Web viewresponsibilities to the division, and to abolish the department

ETHEM ALPAYDIN © The MIT Press, 2010 [email protected] ethem/i2ml2e Lecture Slides for

4641 VOLUME CXLV NUMBER 164 THURSDAY, AUGUST 23, … · VOLUME CXLV NUMBER 164 THURSDAY, AUGUST 23, 2018 Price: $4.00 4641 TABLE OF CONTENTS PUBLIC HEARINGS AND MEETINGS Borough President

INTRODUCTION TO MACHINE LEARNING 3RD EDITION ETHEM ALPAYDINModified by Prof. Carolina Ruiz © The MIT Press, 2014for CS539 Machine Learning at WPI [email protected]

INTRODUCTION TO Machine Learning Adapted from: ETHEM ALPAYDIN Lecture Slides for

INTRODUCTION TO Machine Learningethem/i2ml/slides/v1-1/i2...INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 [email protected] ethem/i2ml Lecture Slides for

Pattern Recognition and its application to image processing · • Neural networks for Pattern Recognition :-Christopher Bishop • Introduction to Machine Learning :- Alpaydin

INTRODUCTION TO Machine Learningusers.cis.fiu.edu/~jabobadi/CAP5610/slides8.pdfINTRODUCTION TO Machine Learning 2nd Edition ETHEM ALPAYDIN, modified by Leonardo Bobadilla and some

CHAPTER 7: Clustering Eick: K-Means and EM (modified Alpaydin transparencies and new transparencies added) Last updated: February 25, 2014

MACHINE LEARNING 8. Clustering. Motivation Based on E ALPAYDIN 2004 Introduction to Machine Learning © The MIT Press (V1.1) 2 Classification problem:

Research Paper Novel Metabolomics Serum Biomarkers for ...4641 Journal of Cancer 2020; 11(16): 4641-4651. doi: 10.7150/jca.41250 Research Paper Novel Metabolomics Serum Biomarkers

INTRODUCTION TO Machine Learning - Rutgers Universityelgammal/classes/cs536/lectures/...1 INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 Edited for CS 536 Fall

INTRODUCTION TO Machine Learningusers.cis.fiu.edu/~jabobadi/CAP5610/slides10.pdfINTRODUCTION TO Machine Learning 2nd Edition ETHEM ALPAYDIN, modified by Leonardo Bobadilla ... (V1.1)

INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 [email protected] ethem/i2ml Lecture Slides for

1 PCA, LDA, HLDA and HDA Reference: 1.E. Alpaydin, “Introduction to Machine Learning,” The MIT Press, 2004. 2.S. R. Searle, “Matrix Algebra Useful for

AMENDMENT TO H.R. 4641 - docs.house.govdocs.house.gov/meetings/IF/IF14/20160420/104854/BILLS-114-4641-E... · AMENDMENT TO H.R. 4641 OFFERED BY MR.ENGEL OF NEW YORK At the end of

Introduction to Machine Learningpattern-recognition.weebly.com/.../machine_learning-alpaydin-2nd_ch… · Introduction to machine learning / Ethem Alpaydin. — 2nd ed. p. cm. Includes

CHAPTER 8: Nonparametric Methods Alpaydin transparencies significantly modified, extended and changed…

INTRODUCTION TO Machine Learning · INTRODUCTION TO Machine Learning ETHEM ALPAYDIN © The MIT Press, 2004 [email protected] ethem/i2ml Lecture Slides for

Lecture 01. Course Overview · 2021. 2. 9. · Lecture 01. Course Overview Nakul Gopalan Post-doctoral Fellow, College of Computing, Georgia Tech Machine Learning CS 4641 B This course

Random Number Generation & Testing Juan Soto [email protected] 301/975-4641

ETHEM ALPAYDIN © The MIT Press, 2010 - BOUN

Multivariate Methods Slides from Machine Learning by Ethem Alpaydin Expanded by some slides from Gutierrez-Osuna