Pattern Recognition: An Overview: Prof. Richard Zanibbi
Pattern Recognition: An Overview: Prof. Richard Zanibbi
An Overview
Prof. Richard Zanibbi
Pattern Recognition
(One) Definition
The identification of implicit objects, types or relationships in raw
data by an animal or machine
Common Problems
• What is it?
• Where is it?
• How is it constructed?
FIGURE 1.1. The objects to be classified are first sensed by a transducer (camera),
whose signals are preprocessed. Next the features are extracted and finally the clas- 8
sification is emitted, here either “salmon” or “sea bass.” Although the information flow
is often chosen to be from the source to the classifier, some systems employ information
flow in which earlier levels of processing can be altered based on the tentative or pre-
liminary response in later levels (gray arrows). Yet others combine two or more stages
Designing a classifier or
clustering algorithm
Feature Selection
Choosing from available features those to be used in our classification
model. Ideally, these:
Feature Extraction
Computing features for inputs at run-time
Preprocessing
User to reduce data complexity and/or variation, and applied before
feature extraction to permit/simplify feature computations; sometimes
involves other PR algorithms (e.g. segmentation)
10
Types of Features
(ordered)
(unordered)
13
A Combination of Features:
Lightness and Width
width
Feature Space
22 salmon sea bass
21 Is now two-
20 dimensional; fish
19 described in model
18 input by a feature vector
17 (x1, x2) representing a
16
point in this space
15
14
2 4 6 8 10
lightness Decision Boundary
FIGURE 1.4. The two features of lightness and width for sea bass and salmon. The dark
A linear discriminant
line could serve as a decision boundary of our classifier. Overall classification error on
In general, determining appropriate features is a
the data shown is lower than if we use only one feature as in Fig. 1.3, but there will(line used to separate
still bedifficult problem,
some errors. and O.
From: Richard determining
Duda, Peter E.optimal
Hart, and features is Pattern
David G. Stork,
c 2001 by John Wiley & Sons, Inc. classes) is shown; still
often
Classification impractical
. Copyright ! or impossible (requires testing
some errors
all feature combinations)
14
Classifier: A Formal Definition
width
22 salmon sea bass
21
20
19
18
17
16
15
14 lightness
2 4 6 8 10
(from Kuncheva: visualizes
FIGURE 1.4. The two features of lightness and width for sea bass and salmon. The dark
changes
line could serve as a decision boundary of our classifier. Overall classification error (gradient) for class score)
on
the data shown is lower than if we use only one feature as in Fig. 1.3, but there will
still be some errors. From: Richard O. Duda, Peter E. Hart, and David G. Stork, Pattern
c 2001 by John Wiley & Sons, Inc.
Classification. Copyright !
18
“Generative” “Discriminative”
Models Models
RE 1.5. Overly complex models for the fish will lead to decision boundaries that
mplicated. While such a decision may lead to perfect classification of our training
es, it would lead to poor performance on future patterns. The novel test point
21
d ? is evidently most likely a salmon, whereas the complex decision boundary
n leads it to be classified as a sea bass. From: Richard O. Duda, Peter E. Hart, and
G. Stork, Pattern Classification. Copyright !c 2001 by John Wiley & Sons, Inc.
Avoiding Over-Fitting
A Trade-off
We may need to accept more errors on our
training set to produce fewer errors on new data
• We have to do this without “peeking at” (repeatedly
evaluating) the test set, otherwise we over-fit the test
set instead
width
22 salmon sea bass
21
20
19
18
17
16
15
14 lightness
2 4 6 8 10
FIGURE 1.6. The decision boundary shown might represent the optimal tradeoff be-
tween performance on the training set and simplicity of classifier, thereby giving the
highest accuracy on new patterns. From: Richard O. Duda, Peter E. Hart, and David 23 G.
Non-
Hierarchical:
all points
assigned to a
cluster each
iteration