Machine Learning

Berlin Chen 2006

References:
1. E. Alpaydin, Introduction to Machine Learning, Chapter 1
2. Kantard, Data Mining: Concepts, Models, Methods and Algorithms, Chapter 1
3. Mitchell, Machine Learning, Chapter 1
4. Han and Kamber, Data Mining: Concepts and Techniques, Chapter 1
 Main Textbooks

1. Ethem Alpaydin, Introduction to Machine Learning,

MIT Press, 2004
http://www.cmpe.boun.edu.tr/~ethem/i2ml/
2. Tom M. Mitchell, Machine Learning, McGraw-Hill,
1997

3. Mehmed M. Kantard, Data Mining: Concepts,

Models, Methods and Algorithms, Wiley-IEEE Press,
2002

4. T. Hastie, R. Tibshirani, and J. Friedman, The

Elements of Statistical Learning; Data Mining,
Inference, and Prediction, Springer-Verlag, 2001

MLDM-Berlin Chen 2
 Reference Textbooks

1. Jiawei Han and Micheline Kamber, Data Mining: Concepts and

Techniques, Morgan Kaufmann, 2001

2. Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern

Approach, Prentice-Hall, 2003

3. Nils J. Nilsson, Artificial Intelligence: A New Synthesis, Morgan

Kaufmann, 1998

4. Baeza-Yates and B. Ribeiro-Neto, Modern Information Retrieval,

Addison Wesley Longman, 1999

5. I. H. Witten and E. Frank, Data Mining, Morgan Kaufmann, 2000.

MLDM-Berlin Chen 3
 Goals

• Know the basic concepts and fundamentals of machine

learning and data mining

• Theoretically understand a variety of models and

algorithms that can be employed in the fields such as
data mining, information retrieval, pattern recognition,
speech processing, image processing, …

MLDM-Berlin Chen 4
 Machine Learning

• Address the question of how to build computer programs

that improve their performance at some task through
experience
– Learning is a process → algorithm/program
• Can be viewed as searching a very large space of
possible hypotheses to determine one that best fits the
observed data and any prior knowledge held by the
learner, and also can correctly generalize to unseen
examples
– Search strategies
– Underlying structures of the hypothesis space

Different learning methods searching different hypothesis spaces

MLDM-Berlin Chen 5
 Machine Learning: Why?

• Recent progress in algorithms and theory

• Growing flood of online data

• Computational power is available

• Budding industry

MLDM-Berlin Chen 6
 Machine Learning: When ?

• Human expertise does not exist (navigating on Mars),

• Humans are unable to explain their expertise (speech

recognition)

• Solution changes in time (routing on a computer network)

• Solution needs to be adapted to particular cases (user

biometrics)

MLDM-Berlin Chen 7
 Machine Learning: Applications

• Association

• Supervised Learning
– Classification
– Regression

• Unsupervised Learning

• Reinforcement Learning

MLDM-Berlin Chen 8
 Associations
• Example: Basket Analysis
– Customer transaction to consumer behavior
• P (Y | X ) probability that somebody who buys X also buys Y
where X and Y are products/services
• Examples:
P( buying “chips” | buying “beer”) = ?
P( buying “Pattern Classification”| buying “Machine Learning” ) = ?

MLDM-Berlin Chen 9
 Classification (1/2)
• Also known as Pattern Recognition
• Example 1: Credit Scoring
– Differentiating 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

MLDM-Berlin Chen 10
 Classification (2/2)

• Face Recognition
Training examples of a person

Test images

AT&T Laboratories, Cambridge UK

http://www.uk.research.att.com/facedatabase.html
MLDM-Berlin Chen 11
 Regression (1/2)

• Example 1: Price of a Used Car

– x : car attributes
– y : price
– y = g (x | θ )
– g (‧): model
– Θ: parameters

MLDM-Berlin Chen 12
 Regression (2/2)

• Example 2: Navigating a Car

– CMU Project: Driving Autonomous Vehicles (1989)
• Autonomous Land Vehicle In a Neural Network (ALVINN)
– Control the angle of the steering wheel, etc.

MLDM-Berlin Chen 13
 Clustering

• Clustering :A kind of unsupervised Learning

• Grouping similar instances

MLDM-Berlin Chen 14
 Reinforcement Learning

• Learning a policy: A sequence of outputs

• No supervised output but delayed reward
• Credit assignment problem
• Game playing
• Robot in a maze
• Multiple agents, partial observability

An action is good if it is part of a good game playing policy

(Is it on the way to win?)

MLDM-Berlin Chen 15
 Other Possible Applications

• Business Management

• Production Control

• Scientific/Medical Research

• …

MLDM-Berlin Chen 16
 Some Issues in Machine Learning

• What algorithms can approximate functions well

(and when) ?
• How does number of training examples influence
accuracy ? (Variance)
• How does complexity of hypothesis representation
impact it ? (Bias)
• How can prior knowledge of learner help ?
• How does noisy data influence accuracy ?
• What are the theoretical limits of learnability ?
• How can systems alter their own representations ?
MLDM-Berlin Chen 17
 What is Data Mining ? (1/4)
• Also called Knowledge Discovery in Databases (KDD),
Information Extraction (IE), Knowledge Extraction (KE) ..

• Emerged during the late 1980s, has made great strides

during the 1990s, and continues to flourish into the new
millennium

• Data-Mining? Information-Mining? Knowledge-Mining?

– Cf. gold mining (but not rock or sand mining)
– The term “Data Mining” is just a misnomer ?

MLDM-Berlin Chen 18
 What is Data Mining ? (2/4)

• Automated data collection tools and mature database

technology lead to tremendous amounts of data stored in
databases, data warehouses and other information
repositories
– Extract/Mine interesting information or knowledge (rules,
regularities, patterns, constraints) from huge amounts of data
stored in databases, data warehouse, and other information
repositories
• Explore hidden and nontrivial facts

– “knowledge mining” from data

MLDM-Berlin Chen 19
 What is Data Mining ? (3/4)

Data Mining Data Mining

Data Information Knowledge

Data Tombs ? Golden Nuggets ?

MLDM-Berlin Chen 20
 What is Data Mining ? (4/4)
• Data mining is an essential step in knowledge discovery

Data Analysis
Data Understanding

Data Cleansing
Data Integration

MLDM-Berlin Chen 21
 Categories of Data Mining

• Predictive Data Mining

– Produce the model of the system described by the given data set
– I.e., perform inference on the current data to make predictions
• Classification
• Regression

• Descriptive Data Mining

– Produce new, nontrivial information (uncover patterns and
relationships) based on the available data set
– Namely, characterize the general properties of the data
• Clustering
• Summarization, or Concept/Class Interpretation
• Dependency/Association Modeling age(X , "20...29") ∧ Income(X ,"20K...29K") ⇒ Buy(X , "CD Player")
• Change and Deviation Detection evolution, outlier detection

MLDM-Berlin Chen 22
 Multi-Dimensional View of Data Mining
• Databases to be mined
– Relational, transactional, object-oriented, object-relational, active,
spatial, time-series, text, multi-media, heterogeneous, legacy,
WWW, etc.
• Knowledge to be mined
– Characterization, discrimination, association, classification,
clustering, trend, deviation and outlier analysis, etc.
– Granularity: mining at multiple levels of abstraction
• Techniques utilized
– Machine learning, statistics, visualization, neural network,
database-oriented, data warehouse (OLAP), etc.
• Applications adapted
– Retail, telecommunication, banking, fraud analysis, DNA mining,
stock market analysis, Web mining, Weblog analysis, etc.

MLDM-Berlin Chen 23
 Roots of Data Mining (1/2)

• Statistics, Mathematics
– Models

• Machine Learning
– Algorithms

• Control theory
– System identification

MLDM-Berlin Chen 24
 Roots of Data Mining (2/2)

• System Identification (an iterative process)

– Structure Identification
– Parameter Identification

u Target system to be identified y

y* _ + predict system’s
Mathematical model y*=f(u,t) ∑ behaviors

y-y*
Identification techniques

Linear ? Nonlinear ?
MLDM-Berlin Chen 25
 Phases of Data Mining (1/7)

1. State the Problem and Formulate the Hypothesis

– The problem statement should be established based on
domain-specific knowledge and experience

– But application studies tend to focus on the data-mining

technique at the expanse of a clear problem statement

– Cooperation between data-mining expertise and application

expertise

experience and treatment

MLDM-Berlin Chen 26
 Phases of Data Mining (2/7)

2. Collect the Data

– Two possible approaches
• Designed experiment
– Data generation process is under control of an expert
• Observational approach (random data generation)
– The expert can not influence the data generation process

– A prior knowledge can be very useful for modeling and final

interpretation of results

– Data respective for estimating a model and testing should come

from the same, unknown, sampling distribution

MLDM-Berlin Chen 27
 Phases of Data Mining (3/7)

3. Preprocessing the Data

– Two tasks involved
• Outlier detection (and removal)
– Outliers are unusual data values that are not consistent
with most observations which can seriously affect
modeling accuracy
– Two strategies for dealing with outliers
» Removal of outliers
» Robust modeling methods

• Scaling (normalization), encoding (discretization), and

selecting features (dimensionality reduction)

– The prior knowledge of application domain should be considered

in data-preprocessing steps
MLDM-Berlin Chen 28
 Phases of Data Mining (4/7)

• Clusters and Outliers

MLDM-Berlin Chen 29
 Phases of Data Mining (5/7)

4. Estimate the Model

– Select and implement the appropriate data-mining technique
• The implementation is based on several models

– Use the technique to learn and discovery information from large

volumes of data sets

MLDM-Berlin Chen 30
 Phases of Data Mining (6/7)

5. Interpret the Model and Draw Conclusions

– Data-mining models should help in decision making

– Data-mining models thus should be interpretable

– Tradeoff between accuracy of model and accuracy of model’s

interpretation

MLDM-Berlin Chen 31
 Phases of Data Mining (7/7)

• All phases and the entire data-mining process are highly

iterative

State the problem

Collect the data

Perform preprocessing

Estimate the model (mine the data)

Interpret the model & draw the conclusions

MLDM-Berlin Chen 32
 Large Data Sets (1/2)

• An exponential growth in information sources and

information-storage units

– The number of hosts are directly proportional to the amount of

data stored on the Internet

MLDM-Berlin Chen 33
 Large Data Sets (2/2)

• Infer knowledge form huge volumes of raw datasets

– Big data can lead to much stronger conclusions

– A rapidly widening gap between data-collection and data-

organization capabilities and the ability to analyze the data

– Manual analysis and semiautomatic computer-based analysis

can not deal with the large volumes of data sets

• Data as the sources for data mining can be classified

into structured, semi-structured and unstructured data
– Traditional data: structured data
– Nontraditional data (multimedia): semi-structured and
unstructured data
MLDM-Berlin Chen 34
Structured Data

MLDM-Berlin Chen 35
 Data Warehouse (1/5)

• Definition
– A collection of integrated, subject-oriented databases designed to
support the decision-support functions (DSF), where each unit of
data is relevant to some moment in time
• Modeled as a multidimensional database structure

– Or, a repository of multiple heterogeneous data sources,

organized under a unified schema usually at a single site in order
to facilitate management decision making

• That is, the sole of a data warehouse is to provide

information for end users for decision support

• Cf. data mart

– A department subset of a data warehouse
MLDM-Berlin Chen 36
Data Warehouse (2/5)

MLDM-Berlin Chen 37
Data Warehouse (3/5)

with OLAP tools

MLDM-Berlin Chen 38
 Data Warehouse (4/5)
• Applications
– Data mining
• Represent one of the major applications for data warehouse
• Provide end-user with the capability to extract hidden,
nontrivial (not obvious) information
– Act as exploratory queries

– Structured query languages (SQL)

• A standard database language
• Used when we know exactly what we are looking for and we
can describe it formally

– Online Analytical Processing (OLAP)

• Do not learn from data, nor create new knowledge
• Let users analyze data by providing multiple views of the
data
MLDM-Berlin Chen 39
 Data Warehouse (5/5)

• Classification of data stored in a data warehouse

– Old detail data
– Current (New) detail data
– Lightly summarized data
– Highly summarized data
– Metadata (the data directory or guide)

• Fundamental types of data transformation

– Simple transformations (encoding/decoding)
– Cleansing and scrubbing
– Integration
– Aggregation and summarization

MLDM-Berlin Chen 40
 A Typical Data Mining System

• Architecture

MLDM-Berlin Chen 41
Confluence of Multiple Disciplines

Machine
Learning
Artificial Neural
Intelligence Networks

Database

Data Mining Information

Retrieval

Statistics

Pattern Knowledge
Recognition Acquisition
Data Visualization/
Knowledge Representation

MLDM-Berlin Chen 42
Topic List and Tentative Schedule

MLDM-Berlin Chen 43
 Resources: Journals

• Journal of Machine Learning Research

• Machine Learning
• Neural Computation
• Neural Networks
• IEEE Transactions on Neural Networks
• IEEE Transactions on Pattern Analysis and Machine
Intelligence
• Annals of Statistics
• Journal of the American Statistical Association
• ...

MLDM-Berlin Chen 44
 Resources: Conferences

• International Conference on Machine Learning (ICML)

• European Conference on Machine Learning (ECML)
• Neural Information Processing Systems (NIPS)
• Uncertainty in Artificial Intelligence (UAI)
• Computational Learning Theory (COLT)
• International Joint Conference on Artificial Intelligence
(IJCAI)
• International Conference on Neural Networks

MLDM-Berlin Chen 45