Introduction to Pattern Recognition

Selim Aksoy
Department of Computer Engineering
Bilkent University
[email protected]

CS 551, Spring 2014

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

1 / 40

Human Perception

Humans have developed highly sophisticated skills for

sensing their environment and taking actions according to
what they observe, e.g.,
I
I
I
I

recognizing a face,
understanding spoken words,
reading handwriting,
distinguishing fresh food from its smell.

We would like to give similar capabilities to machines.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

2 / 40

What is Pattern Recognition?

I

A pattern is an entity, vaguely defined, that could be given a

name, e.g.,
I
I
I
I
I
I

fingerprint image,
handwritten word,
human face,
speech signal,
DNA sequence,
...

Pattern recognition is the study of how machines can

I
I
I

observe the environment,

learn to distinguish patterns of interest,
make sound and reasonable decisions about the categories
of the patterns.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

3 / 40

Human and Machine Perception

I

We are often influenced by the knowledge of how patterns

are modeled and recognized in nature when we develop
pattern recognition algorithms.

Research on machine perception also helps us gain deeper

understanding and appreciation for pattern recognition
systems in nature.

Yet, we also apply many techniques that are purely

numerical and do not have any correspondence in natural
systems.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

4 / 40

Pattern Recognition Applications

Table 1: Example pattern recognition applications.
Problem Domain

Application

Input Pattern

Pattern Classes

Document image analysis

Document classification
Document classification
Multimedia database retrieval
Speech recognition

Optical character recognition

Internet search
Junk mail filtering
Internet search
Telephone directory assistance
Information extraction
Personal identification

Document image
Text document
Email
Video clip
Speech waveform

Characters, words
Semantic categories
Junk/non-junk
Video genres
Spoken words

Natural language processing

Biometric recognition
Medical
Military
Industrial automation
Industrial automation
Remote sensing
Bioinformatics
Data mining

CS 551, Spring 2014

Sentences
Face, iris, fingerprint

Parts of speech
Authorized users for access
control
Computer aided diagnosis
Microscopic image
Cancerous/healthy cell
Automatic target recognition Optical or infrared image
Target type
Printed circuit board inspec- Intensity or range image
Defective/non-defective prodtion
uct
Fruit sorting
Images taken on a conveyor Grade of quality
belt
Forecasting crop yield
Multispectral image
Land use categories
Sequence analysis
DNA sequence
Known types of genes
Searching for meaningful pat- Points in multidimensional Compact and well-separated
terns
space
clusters

c
2014,
Selim Aksoy (Bilkent University)

5 / 40

Pattern Recognition Applications

Figure 1: English handwriting recognition.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

6 / 40

Pattern Recognition Applications

Figure 2: Chinese handwriting recognition.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

7 / 40

Pattern Recognition Applications

Figure 3: Fingerprint recognition.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

8 / 40

Pattern Recognition Applications

Figure 4: Biometric recognition.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

9 / 40

Pattern Recognition Applications

Figure 5: Autonomous navigation.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

10 / 40

Pattern Recognition Applications

Figure 6: Cancer detection and grading using microscopic tissue data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

11 / 40

Pattern Recognition Applications

Figure 7: Cancer detection and grading using microscopic tissue data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

12 / 40

Pattern Recognition Applications

Figure 8: Land cover classification using satellite data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

13 / 40

Pattern Recognition Applications

Figure 9: Building and building group recognition using satellite data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

14 / 40

Pattern Recognition Applications

Figure 10: License plate recognition: US license plates.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

15 / 40

Pattern Recognition Applications

Figure 11: Clustering of microarray data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

16 / 40

An Example
I

Problem: Sorting incoming

fish on a conveyor belt
according to species.

Assume that we have only

two kinds of fish:
I
I

sea bass,
salmon.

Figure 12: Picture taken from a

camera.
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

17 / 40

An Example: Decision Process

I

What kind of information can distinguish one species from

the other?
I

What can cause problems during sensing?

I

length, width, weight, number and shape of fins, tail shape,

etc.

lighting conditions, position of fish on the conveyor belt,

camera noise, etc.

What are the steps in the process?

I

capture image isolate fish take measurements make

decision

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

18 / 40

An Example: Selecting Features

Assume a fisherman told us that a sea bass is generally

longer than a salmon.

We can use length as a feature and decide between sea

bass and salmon according to a threshold on length.

How can we choose this threshold?

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

19 / 40

An Example: Selecting Features

Figure 13: Histograms of the length feature for two types of fish in training
samples. How can we choose the threshold l to make a reliable decision?
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

20 / 40

An Example: Selecting Features

Even though sea bass is longer than salmon on the

average, there are many examples of fish where this
observation does not hold.

Try another feature: average lightness of the fish scales.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

21 / 40

An Example: Selecting Features

Figure 14: Histograms of the lightness feature for two types of fish in training
samples. It looks easier to choose the threshold x but we still cannot make a
perfect decision.
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

22 / 40

An Example: Cost of Error

We should also consider costs of different errors we make

in our decisions.

For example, if the fish packing company knows that:

I

Customers who buy salmon will object vigorously if they see

sea bass in their cans.
Customers who buy sea bass will not be unhappy if they
occasionally see some expensive salmon in their cans.

How does this knowledge affect our decision?

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

23 / 40

An Example: Multiple Features

I

Assume we also observed that sea bass are typically wider

than salmon.

We can use two features in our decision:

I
I

lightness: x1
width: x2

Each fish image is now represented as a point (feature

vector )
!
x1
x=
x2
in a two-dimensional feature space.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

24 / 40

An Example: Multiple Features

Figure 15: Scatter plot of lightness and width features for training samples.
We can draw a decision boundary to divide the feature space into two
regions. Does it look better than using only lightness?
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

25 / 40

An Example: Multiple Features

Does adding more features always improve the results?

I
I
I
I

Avoid unreliable features.

Be careful about correlations with existing features.
Be careful about measurement costs.
Be careful about noise in the measurements.

Is there some curse for working in very high dimensions?

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

26 / 40

An Example: Decision Boundaries

I

Can we do better with another decision rule?

More complex models result in more complex boundaries.

Figure 16: We may distinguish training samples perfectly but how can
we predict how well we can generalize to unknown samples?
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

27 / 40

An Example: Decision Boundaries

I

How can we manage the tradeoff between complexity of

decision rules and their performance to unknown samples?

Figure 17: Different criteria lead to different decision boundaries.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

28 / 40

More on Complexity

1

Figure 18: Regression example: plot of 10 sample points for the input
variable x along with the corresponding target variable t. Green curve is the
true function that generated the data.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

29 / 40

More on Complexity



(a) 0th order polynomial



(c) 3rd order polynomial



(b) 1st order polynomial



(d) 9th order polynomial

Figure 19: Polynomial curve fitting: plots of polynomials having various

orders, shown as red curves, fitted to the set of 10 sample points.
CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

30 / 40

More on Complexity



(a) 15 sample points



(b) 100 sample points

Figure 20: Polynomial curve fitting: plots of 9th order polynomials fitted to
15 and 100 sample points.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

31 / 40

Pattern Recognition Systems

Physical environment
Data acquisition/sensing

Training data

Preprocessing

Preprocessing

Feature extraction

Feature extraction/selection

Features

Features

Classification

Model

Model learning/estimation

Postprocessing
Decision

Figure 21: Object/process diagram of a pattern recognition system.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

32 / 40

Pattern Recognition Systems

I

Data acquisition and sensing:

I
I

Pre-processing:
I
I

Measurements of physical variables.

Important issues: bandwidth, resolution, sensitivity,
distortion, SNR, latency, etc.

Removal of noise in data.

Isolation of patterns of interest from the background.

Feature extraction:
I

Finding a new representation in terms of features.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

33 / 40

Pattern Recognition Systems

I

Model learning and estimation:

I

Classification:
I

Learning a mapping between features and pattern groups

and categories.

Using features and learned models to assign a pattern to a

category.

Post-processing:
I
I
I

Evaluation of confidence in decisions.

Exploitation of context to improve performance.
Combination of experts.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

34 / 40

The Design Cycle

Select
features

Collect
data

Select
model

Train
classifier

Evaluate
classifier

Figure 22: The design cycle.

Data collection:
I
I

Collecting training and testing data.

How can we know when we have adequately large and
representative set of samples?

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

35 / 40

The Design Cycle

I

Feature selection:
I
I
I

Domain dependence and prior information.

Computational cost and feasibility.
Discriminative features.
I
I

Similar values for similar patterns.

Different values for different patterns.

Invariant features with respect to translation, rotation and

scale.
Robust features with respect to occlusion, distortion,
deformation, and variations in environment.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

36 / 40

The Design Cycle

I

Model selection:
I
I
I
I
I
I

Domain dependence and prior information.

Definition of design criteria.
Parametric vs. non-parametric models.
Handling of missing features.
Computational complexity.
Types of models: templates, decision-theoretic or statistical,
syntactic or structural, neural, and hybrid.
How can we know how close we are to the true model
underlying the patterns?

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

37 / 40

The Design Cycle

Training:
I
I

How can we learn the rule from data?

Supervised learning: a teacher provides a category label or
cost for each pattern in the training set.
Unsupervised learning: the system forms clusters or natural
groupings of the input patterns.
Reinforcement learning: no desired category is given but the
teacher provides feedback to the system such as the
decision is right or wrong.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

38 / 40

The Design Cycle

Evaluation:
I

I
I

How can we estimate the performance with training

samples?
How can we predict the performance with future data?
Problems of overfitting and generalization.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

39 / 40

Summary
I

Pattern recognition techniques find applications in many

areas: machine learning, statistics, mathematics, computer
science, biology, etc.

There are many sub-problems in the design process.

Many of these problems can indeed be solved.

More complex learning, searching and optimization

algorithms are developed with advances in computer
technology.

There remain many fascinating unsolved problems.

CS 551, Spring 2014

c
2014,
Selim Aksoy (Bilkent University)

40 / 40