csci 5521: pattern recognition - university of minnesota
TRANSCRIPT
Prof. Paul Schrater Pattern Recognition CSCI 5521 2
Business
• Check to make sure you received the test email. Ifnot, you are not officially registered.
• Course web page:http://gandalf.psych.umn.edu/~schrater/schrater_lab/courses/PattRecog03/PattRecog.html
Prof. Paul Schrater Pattern Recognition CSCI 5521 3
Syllabus• Reading materials:
– Statistical Pattern Recognition, 2nd Ed. Andrew Webb– Pattern Classification, 2nd Ed. Duda, Hart, Stork ( Select
Chapters will be posted for download)– Neural Networks for Pattern Recognition. Bishop ( Select
Chapters will be posted for download)– Papers posted on the web site.– Downloads will be password protected.
• Grading• 50% on the homework assignments• 20% on the midterm• 30% on the final project. !
Prof. Paul Schrater Pattern Recognition CSCI 5521 4
Syllabus cont’d• Final Project 10-15 page paper involving:1) Simulation or experiments. For example, implement a pattern
recognition system for a particular application, e.g. digitclassification, document clustering, etc.
2) Literature survey (with critical evaluation) on a given topic.3) Theoretical work (detailed derivations, extensions of existing
work, etc)
Important dates:• Sept. 23: Topic selection. One or two pages explaining the
project with a list of references.• Nov. 4: Partial report (3 to 5 pages).• Dec. 16: Final report (10 to 15 pages).• Students may work in groups of 2-3.
Prof. Paul Schrater Pattern Recognition CSCI 5521 5
Policies/Procedures
• DO NOT CHEAT.• Do NOT work in groups for homework• Electronically submit homework.• Homework must be submitted before class
on the day it is due.
Introduction to PatternRecognition
• Syllabus• What are Patterns?• Pattern Recognition• An Example• Pattern Recognition Systems• The Design Cycle• Learning and Adaptation• Conclusion
Prof. Paul Schrater Pattern Recognition CSCI 5521 8
Examples of Patterns
Natural orNot?
How can wedescribe thesepatterns?
Prof. Paul Schrater Pattern Recognition CSCI 5521 9
Shape Patterns
D’arcy Thompson’s suggestion ofspecies change throughcontinuous deformation
This figure shows the effects ofAlzheimer's Disease on the ventricularexpansion rate measured from serial MRI.
Prof. Paul Schrater Pattern Recognition CSCI 5521 10
Explaining patterns Voice Puppetry, M. Brand;
Siggraph’99
Prof. Paul Schrater Pattern Recognition CSCI 5521 13
What is a Pattern?• A set of instances that:
– Share some regularities and similarities.– Are Repeatable.– Are Observable, sometimes partially, using sensors
with noise and distortions.• How do we define “regularity”?• How do we define “similarity”?• How do we define “likelihood” for the repetition
of a pattern?• How do we model the sensors?• What is not a pattern?
Prof. Paul Schrater Pattern Recognition CSCI 5521 14
Two Schools of Pattern Rec.• Generative methods:
Bayesian school, pattern theory.1). Define patterns and regularities (graph spaces),2). Specify likelihood model for how signals are
generated from hidden structures3). Learning probability models from ensemble of signals4). Inferences.
• Discriminative methods:– The goal is to tell apart a number of patterns, say 100
people, 10 digits, directly, without understanding ormathematical description.
– “You should not solve a problem to an extend morethan what you need”.
Prof. Paul Schrater Pattern Recognition CSCI 5521 15
Pattern Recognition applications
• Build a machine that can recognize patterns:– Speech recognition
– Fingerprint identification
– OCR (Optical Character Recognition)
– DNA sequence identification
Prof. Paul Schrater Pattern Recognition CSCI 5521 16
An Example
• “Sorting incoming Fish on a conveyoraccording to species using optical sensing”
Sea bassSpecies
Salmon
Prof. Paul Schrater Pattern Recognition CSCI 5521 17
• Problem Analysis
– Set up a camera and take some sample images to extract features
• Length• Lightness• Width• Number and shape of fins• Position of the mouth, etc…
• This is the set of all suggested features to explore for use in our classifier!
Prof. Paul Schrater Pattern Recognition CSCI 5521 18
PreprocessingSegment fish fromBackground
Feature Extraction Image data from each fish Summarized by feature extractor whose purpose is to reduce the data by measuring certain features
ClassificationThe features are passedto a classifier, that uses thefeatures to decide whichclass the instance belongsto.
Prof. Paul Schrater Pattern Recognition CSCI 5521 20
The length is a poor feature alone!
Select the lightness as a possiblefeature.
Prof. Paul Schrater Pattern Recognition CSCI 5521 22
• Threshold decision boundary and cost relationship
– Move our decision boundary toward smaller values oflightness in order to minimize the cost (reduce thenumber of sea bass that are classified salmon!)
Task of decision theory
Prof. Paul Schrater Pattern Recognition CSCI 5521 23
• Adopt the lightness and add the width of thefish
Fish xT = [x1, x2]
Lightness Width
Prof. Paul Schrater Pattern Recognition CSCI 5521 25
• We might add other features that are not correlatedwith the ones we already have. A precaution shouldbe taken not to reduce the performance by addingsuch “noisy features”
• Ideally, the best decision boundary should be theone which provides an optimal performance such asin the following figure:
Prof. Paul Schrater Pattern Recognition CSCI 5521 27
• However, our satisfaction is premature becausethe central aim of designing a classifier is tocorrectly classify novel input
Issue of generalization!
Prof. Paul Schrater Pattern Recognition CSCI 5521 29
Pattern Recognition Systems
• Sensing
– Use of a transducer (camera or microphone)– PR system depends of the bandwidth, the resolution
sensitivity distortion of the transducer
• Segmentation and grouping
– Patterns should be well separated and should notoverlap
Prof. Paul Schrater Pattern Recognition CSCI 5521 31
• Feature extraction– Discriminative features– Invariant features with respect to translation, rotation and scale.
• Classification– Use a feature vector provided by a feature extractor to assign the
object to a category
• Post Processing– Exploit context input dependent information other than from the
target pattern itself to improve performance
Prof. Paul Schrater Pattern Recognition CSCI 5521 32
The Design Cycle
• Data collection• Feature Choice• Model Choice• Training• Evaluation• Computational Complexity
Prof. Paul Schrater Pattern Recognition CSCI 5521 34
• Data Collection
– How do we know when we have collected anadequately large and representative set ofexamples for training and testing the system?
Prof. Paul Schrater Pattern Recognition CSCI 5521 35
• Feature Choice
– Depends on the characteristics of the problemdomain. Simple to extract, invariant toirrelevant transformation insensitive to noise.
Prof. Paul Schrater Pattern Recognition CSCI 5521 36
• Model Choice
– Unsatisfied with the performance of our fishclassifier and want to jump to another class ofmodel
Prof. Paul Schrater Pattern Recognition CSCI 5521 37
• Training
– Use data to determine the classifier. Manydifferent procedures for training classifiers andchoosing models
Prof. Paul Schrater Pattern Recognition CSCI 5521 38
• Evaluation
– Measure the error rate (or performance andswitch from one set of features to another one
Prof. Paul Schrater Pattern Recognition CSCI 5521 39
• Computational Complexity
– What is the trade-off between computationalease and performance?
– (How an algorithm scales as a function of thenumber of features, patterns or categories?)
Prof. Paul Schrater Pattern Recognition CSCI 5521 40
Learning and Adaptation
• Supervised learning
– A teacher provides a category label or cost for eachpattern in the training set
• Unsupervised learning
– The system forms clusters or “natural groupings” of theinput patterns