Beruflich Dokumente
Kultur Dokumente
()
Instructor
MinLingZhang()
Email:zhangml@seu.edu.cn
URL:http://cse.seu.edu.cn/PersonalPage/zhangml/
SoutheastUniversity
Soochow,FallSemester
Textbook
Richard O. Duda, Peter E. Hart,
David G. Stork
Pattern Classification,
2nd edition
(2)
, 2004
WebResources
InternationalAssociationforPatternRecognition(IAPR)
PatternRecognitionJournal(PRJ)
Listofpatternrecognitionwebsites
Carefullyreadand
Ourcourseisnt
comprehendmaterialsin
amathematical
one AppendixMathematical
Foundation
TerminologiesandContents
Importantanddifficult
oneswillbeannotated Onlyforreference
andevenrevisited purpose
withChinese
Pattern Recognition Soochow,FallSemester 8
Chapter 1
Introduction
Therearevariouskindsofpatterns
Visualpatterns()suchaseyes,nose,
mouth,face,fingerprint,etc.
Temporalpatterns()suchasspeech,
audios,videos,datastreams,etc.
Logicalpatterns()suchascharacters,
strings,images,etc.
Wealreadyknowthecategories
ofcharacters,andthenclassify
thehandwrittenonesinto
categoryAandcategoryB
Clustering: An example
Wedonotknowthecategories
ofsymbols,andthenlearnthe
categoriesandgroupthe
symbolsaccordingly
()
Itisthestudyofhowmachinescan
Perceive:Observetheenvironment(i.e.
interactwiththerealworld)
Process:Learn todistinguishpatternsof
interestfromtheirbackground
Prediction: make soundandreasonable
decisionsaboutthecategoriesofthepatterns
Theproblemofsearchingforpatternsindataisa
fundamentaloneandhasalongandsuccessfulhistory.
ChristopheM.Bishop
Pattern recognition is needed in designing
almost all automated and intelligent systems!
Usefulinscenariossuchasautomaticlicenseplate
recognition(ALPR), opticalcharacterrecognition
(OCR),etc.
Usefulinscenariossuchas
speechtotext(STT), voice
command&control, etc.
Usefulinscenariossuchas
computerized accesscontrol,
criminalpursuit,etc.
Usefulinscenariossuchas
digitalsignatureverification,
creditcardantifraud, etc.
Input: Output:
imageswithseveral locationsofthepeoples
people facesintheimage
Usefulinscenariossuchas
informationretrieval,
documentorganization, etc.
Forcomputers,patternrecognitionisnevereasy
All in all, pattern recognition is important,
useful, attractive, but rather challenging
Challenges Opportunities
Feature()
Attributeswhichcharacterizepropertiesofthesamples
[]
e.g. to characterize a person, we may use features such
as height, weight, age, salary, occupation, etc.
vector component
(initalic)
transpose
operator
featurevector
(inboldface) dimensionality
(numberoffeatures)
ScatterPlot()
Eachsampleisplottedasa
pointinthefeaturespace
(
)
a 2D scatter plot
Separateseabass
fromsalmon
Threebasic
[ vs.] steps
Adjustthelevelofillumination
Routineimage
processing Denoising
Enhancethelevelofcontrast
Isolatedifferentfishesfromone
segmentation another
Isolatefishesfromthebackground
......
Seabassis lengthcouldbeagoodcandidate
usuallylonger
than asalmon forfeatures
......
haxis:lengthoffish
vaxis:numberoffisheswith
acertainlength
Onaverage,seabassis
somewhat longerthansalmon
Toomuchoverlaps
poorseparationwith
thelengthfeature
histogramforlength
Onaverage,seabassis
much brighterthansalmon
Lessoverlaps better
separationwiththelightness
feature,butstillabit
histogramforlightness unsatisfactory
blackdots:salmonsamples
reddots:seabasssamples
Linear decisionboundary:
Complex decision
boundary
Performance on
the training set
Tradeoff
Simplicity of
the classifier
Associative Memory ()
Pattern Pattern preliminary steps in
Regression ()
pattern recognition
Apostprocessordecideontheappropriate
Inadditionto actionbasedontheclassification
theusual
bottomup Aclassifierusesextractedfeaturesto
flowofdata, assignthesensedobjecttoacategory
somesystems
alsoemploy
Afeatureextractormeasuresobject
feedback propertiesthatareusefulforclassification
fromhigher
levelsback Asegmentor isolatessensedobjectsfrom
downtolower thebackgroundorfromotherobject
levels(gray
arrows) Asensorconvertsphysicalinputs(e.g.
images,sounds)intodigitalsignaldata
Whatkindsoffeaturesarepromising?
DistinguishingCapability: Whosevaluesareverysimilarfor
objectsinthesamecategory,whileverydifferentforobjectsin
differentcategories
Whatifalargesetofcandidatefeaturesavailable?
Choosethosearesimpletoextract
Choosethosearerobusttonoise
Choosethosecanleadtosimplerdecisionboundaries
DesiredProperties
Patternsfromthesameclassesshouldhavesimilarrepresentations
Patternsfromdifferentclassesshouldhavedissimilar
representations
Patternrepresentationsshouldbeinvarianttotransformationssuch
astranslations,rotations,resizes,reflections,nonrigiddeformations
Intraclassvariationshouldbesmall
Interclassvariationshouldbelarge
Pattern Recognition Soochow,FallSemester 49
Missing Features
Inpracticalproblems,valuesforcertainfeatures
maybemissing
Occlusionbetweenfishes fishwidthcantbemeasured
Howcouldwetrainclassifierswithmissingfeatures?
Navemethodcouldbeused,butmaynotbeoptimal
Assumingthevalueofmissingfeaturesiszero
Assigningtheaveragevalueofpatternsalreadyseenforthe
missingfeature
Sophisticatedmethodmightbebetter,butrequiresextra
effortsintermsofstorageandtime
Fillinthemissingvalueswithregressiontechniques
Thesamepatternwithindifferentcontextmighthave
differentmeanings
Usethecontextofaconversationtoinferthemeaningofthespeaker
Contextisveryhelpful!
Riskistotalexpectedcostwhichwewanttooptimize
Errorrate(percentagesoftestpatternsbeingwronglyclassified)
Precision,Recall,AreaundertheROCcurve(AUC),etc.
Questionsoncostsandrisks
Howdoweincorporateknowledgeofcosts,e.g.unequalcost?
Canweestimatethelowest possibleriskofanyclassifier?
Pattern Recognition Soochow,FallSemester 55
Computational Complexity
Howdoesanalgorithmscalewith
Thenumberoffeatures(dimensionality)
Thenumberoftrainingpatterns
Thenumberofpossiblecategories
Bruteforce()approachesmightleadtoperfect
classification,butwithimpracticaltimeandstorage
requirements
InOCR,labelallpossible20x20binarypixelimageswithacategory
usesimpletablelookup()toclassifyincomingpatterns
Labelingeachofthe220x20(10120)patternsisprohibitive
Howcanwefindagoodtradeoffbetweencomputational
easeandclassifierperformance?
Pattern Recognition Soochow,FallSemester 56
Summary
WhatisPatternRecognition?
Pattern
Theoppositeofchaos
Variouskinds:visualpatterns,temporalpatterns,logicalpatterns,etc.
Recognition
Identificationofapatternasamemberofacategory
Classification:categoriesknown assignproperclasslabelforeach
pattern
Clustering:categoriesunknown learncategoriesandgrouppatterns
PatternRecognition
Perceive:observetheenvironment(i.e.interactwiththerealworld)
Process:learn todistinguishpatternsofinterest
Prediction:make soundandreasonabledecisionsaboutthecategories