Chapter 4 – Introduction to
Machine Learning
AP Dr Mohamad Hafis Izran Bin Ishak
Control & Mechatronics Engineering Division (CMED)
School of Electrical Engineering
Faculty of Engineering
Universiti Teknologi Malaysia.
SKEM 4173– Artificial Intelligence
[Link]
SKEM 4173 innovative • innovative
entrepreneurial • entrepreneurial • global
• global
Chapter Outline
4. Introduction to Machine Learning
•4.1 Definition of Machine Learning
•4.2 Types of Learning Algorithm
•4.3 Examples of Learning
Techniques
SKEM 4173 innovative • entrepreneurial • global
4.0
Introduction to Machine
Learning
Definition of Machine Learning
Types of Learning Algorithm
Examples of Learning Techniques
SKEM 4173 innovative • entrepreneurial • global 3
Machine Learning is…
Machine learning, a branch of artificial intelligence, concerns
the construction and study of systems that can learn from data.
SKEM 4173 innovative • entrepreneurial • global
Machine Learning is…
Machine learning is programming computers to optimize a
performance criterion using example data or past experience.
-- Ethem Alpaydin
The goal of machine learning is to develop methods that can
automatically detect patterns in data, and then to use the
uncovered patterns to predict future data or other outcomes of
interest.
-- Kevin P. Murphy
The field of pattern recognition is concerned with the automatic
discovery of regularities in data through the use of computer
algorithms and with the use of these regularities to take actions.
-- Christopher M. Bishop
SKEM 4173 innovative • entrepreneurial • global
Machine Learning is…
Machine learning is about predicting the future, based on the
past.
-- Hal Daume III
past future
Training model/ Testing model/
Data predictor Data predictor
SKEM 4173 innovative • entrepreneurial • global
Algorithms
• The success of machine learning system depends on
the algorithms.
• The algorithms control the search to find and build
the knowledge structures.
• The learning algorithms should extract useful
information from training examples.
SKEM 4173 innovative • entrepreneurial • global
Types of Learning Algorithm
1. Supervised (inductive) learning
– Training data includes desired outputs
2. Unsupervised learning
– Training data does not include desired outputs
3. Semi-supervised learning
– Training data includes a few desired outputs
4. Reinforcement learning
– Rewards from sequence of actions
SKEM 4173 innovative • entrepreneurial • global
Algorithms
Supervised learning Unsupervised learning
Semi-supervised learning
9
SKEM 4173 innovative • entrepreneurial • global
• Both supervised and reinforcement learning use mapping between input
and output
• Supervised learning where the feedback provided to the agent is correct
set of actions for performing a task, reinforcement learning uses rewards
and punishments as signals for positive and negative behavior.
SKEM 4173 innovative • entrepreneurial • global
Learning techniques
1. Supervised Learning
• Supervised learning categories and techniques:
– Linear classifier (numerical functions)
– Parametric (Probabilistic functions)
• Naïve Bayes, Gaussian discriminant analysis (GDA), Hidden Markov
models (HMM), Probabilistic graphical models
– Non-parametric (Instance-based functions) LAZY LEARNER
• K-nearest neighbors, Kernel regression, Kernel density estimation,
Local regression
– Non-metric (Symbolic functions)
• Classification and regression tree (CART), decision tree
– Aggregation
• Bagging (bootstrap + aggregation), Adaboost, Random forest
SKEM 4173 innovative • entrepreneurial • global
Supervised Learning
• Linear classifier
• Techniques:
– Perceptron
– Logistic regression
– Support vector machine (SVM) overview here
– Adaline
– Multi-layer perceptron (MLP)
SKEM 4173 innovative • entrepreneurial • global
Data
examples
Data
SKEM 4173 innovative • entrepreneurial • global
Supervised learning
examples
label
label1
label2
labeled examples
label3
label4
Supervised learning: given labeled examples
SKEM 4173 innovative • entrepreneurial • global
Supervised learning
label
label1
model/
label2 predictor
label3
label4
Supervised learning: given labeled examples
SKEM 4173 innovative • entrepreneurial • global
Supervised learning
model/ predicted label
predictor
Supervised learning: learn to predict new example
SKEM 4173 innovative • entrepreneurial • global
Classification Example
Differentiate
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
SKEM 4173 innovative • entrepreneurial • global
Classification Applications
Face recognition
Character recognition
Spam detection
Medical diagnosis: From symptoms to illnesses
Biometrics: Recognition/authentication using physical
and/or behavioral characteristics: Face, iris, signature, etc
...
SKEM 4173 innovative • entrepreneurial • global
Face Recognition
Training examples of a person
Test images
AT&T Laboratories, Cambridge UK
[Link]
SKEM 4173 innovative • entrepreneurial • global 19
Regression Example
Price of a used car
x : car attributes y = wx+w0
(e.g. mileage)
y : price
SKEM 4173 innovative • entrepreneurial • global 20
Regression Applications
Economics/Finance: predict the value of a stock
Epidemiology
Car/plane navigation: angle of the steering wheel,
acceleration, …
Temporal trends: weather over time
SKEM 4173 innovative • entrepreneurial • global
Example: Spam Email Filtering
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
• Linear Support Vector Machine
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
• Ideally the best decision boundary should be the one
which provides an optimal performance such as
follows:
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
• Lazy Learner (Instance based
learning)
SKEM 4173 innovative • entrepreneurial • global
• Lazy Learner (Instance based
learning)
SKEM 4173 innovative • entrepreneurial • global
• What K-NN does?
SKEM 4173 innovative • entrepreneurial • global
• The K-NN steps:
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
• Decision Tree
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
Not only applied for Decision Tree technique
SKEM 4173 innovative • entrepreneurial • global
Underfitting vs Overfitting
SKEM 4173 innovative • entrepreneurial • global
SKEM 4173 innovative • entrepreneurial • global
Learning techniques
2. Unsupervised Learning
• Unsupervised learning categories and techniques
– Clustering
• K-means clustering
• Spectral clustering
– Density Estimation
• Gaussian mixture model (GMM)
• Graphical models
– Dimensionality reduction
• Principal component analysis (PCA)
• Factor analysis
SKEM 4173 innovative • entrepreneurial • global
Unsupervised learning
Unsupervised learning: given data, i.e. examples, but no labels
SKEM 4173 innovative • entrepreneurial • global
Unsupervised learning
applications
learn clusters/groups without any label
customer segmentation (i.e. grouping)
image compression
bioinformatics: learn motifs
…
SKEM 4173 innovative • entrepreneurial • global
Learning techniques
3. Reinforcement Learning (RL)
• Reinforcement Learning(RL) is a type of machine learning
technique that enables an agent to learn in an interactive
environment by trial and error using feedback from its own
actions and experiences.
Reinforcement learning goal is to find a suitable action model
that would maximize the total cumulative reward of the agent.
SKEM 4173 innovative • entrepreneurial • global
Reinforcement Learning
left, right, straight, left, left, left, straight GOOD
left, straight, straight, left, right, straight, straight BAD
left, right, straight, left, left, left, straight 18.5
left, straight, straight, left, right, straight, straight -3
Given a sequence of examples/states and a reward after
completing that sequence, learn to predict the action to take
in for an individual example/state
SKEM 4173 innovative • entrepreneurial • global
RL – Pacman Game
• Goal of the PacMan is to eat the food in the grid while avoiding
the ghosts on its way.
• What is the agent, environment, state, reward and action?
SKEM 4173 innovative • entrepreneurial • global
RL Algorithms
• Markov Decision Processes (MDPs)
• Q-learning Pacman
• SARSA (State-Action-Reward-State-Action)
• Deep Q-Networks (DQNs)
• Deep Deterministic Policy Gradient (DDPG)
• ….
SKEM 4173 innovative • entrepreneurial • global
Modern applications of
artificial intelligence
SKEM 4173 innovative • entrepreneurial • global
End of Chapter 4
SKEM 4173 innovative • entrepreneurial • global 55
