Topic for the class:Well-Posed Learning Problems

Module -1: Title: Introduction

Date & Time : 25 August 2020, 11AM-12PM

Dr. Sireesha Rodda

Professor
Department of Computer Science and Engineering
GITAM Institute of Technology (GIT)
Visakhapatnam – 530045
Email: [email protected]

Wednesday, July 14, 2021 Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning 1
 Course objectives
1. To understand the concepts of Machine Learning - Concept learning
and Decision Trees
2.  To gain an insight into Neural network learning and study of genetic
algorithms.
3. To possess knowledge in Bayesian Learning
4. To explore the concept of Instance-based Learning.
5. To understand the inference and learning algorithms for Hidden Markov
Model.

Wednesday, July 14, 2021 Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning 2
 Learning Outcomes
At the end of the session, the students will be able to:
• Identify tasks in real world which can be improved with
the usage of machine learning techniques
• Discuss the underlying assumptions as well a s issues
inherent to machine learning
• Define a machine learning problem formally as a well
posed learning problem.

Wednesday, July 14, 2021 Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning 3
 Machine Learning

• The field of Machine Learning is concerned with the question of

how to construct computer programs that automatically improve
with experience
• The purpose of this course is to present key algorithms and theory
that form the core of Machine Learning

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Machine Learning

• Interdisciplinary nature of the material:

Statistics, Artificial Intelligence, Information Theory, etc.

• Basic question:
How to program computers to learn?

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Machine Learning

Intelligent Data Analysis:

• Intelligent application of data analytic tools (Statistics)
• Application of “intelligent” data analytic tools (Machine Learning)

Modern world: Data-driven world (industrial, commercial, financial,

scientific activities)

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?

• Recent progress in algorithms and theory

• Growing flood of online data

• Computational power available

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?

• Niches for Machine Learning:

– Data Mining: using historical data to improve decisions
Medical records  medical knowledge
– Software applications we can’t program by hand
Autonomous driving
Speech recognition

– Self customizing programs

Newsreader that learns user interests

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?
• Data Mining
– Data: Recorded facts
– Information: Set of patterns, or expectations, that underlie the data
– Data Mining: Extraction of implicit, previously unknown, and potentially useful
information from data
– Machine Learning: Provides the technical basis of data mining

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?

• Typical Datamining Tasks

– Risk of Emergency Cesarean Section
Given
• 9714 patient records, each describing a pregnancy and birth
• Each patient record contains 215 features
Learn to predict:
• Classes of patients at high risk for emergency cesarean section

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?
– Credit Risk Analysis

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?
– Customer Retention

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Why Machine Learning?

– Problems Too Difficult to Program by Hand

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
What the future holds for Machine Learning

Today: tip of the iceberg

• First-generation algorithms: neural nets, decision trees, regression....
• Applied to well-formated databases
Tomorrow: enormous impact
• Learn across mixed-media data and multiple databases
• Learn by active experimentation
• Learn decisions rather than predictions
• Cumulative, life-long learning

Department of Computer Science and Engineering, GIT Course Code and Course Title:EID 403: Machine Learning
 Learning
• How to program computers to learn?
Learning: Improving automatically with experience
• Example: Computers learning from medical records which treatments
are most effective for new diseases
• Added value: Better understanding of human learning abilities
 Well-posed Learning Problems

1.1 Well-Posed Learning Problems

– Definition:
A computer program is said to learn from experience E with respect to
some class of tasks T and performance measure P, if its performance at
tasks in T, as measured by P, improves with experience E
 An example well-posed learning problem

• Example
– Task T: Playing checkers
– Training Experience E: Playing games against itself
– Performance Measure P: Percentage of games won against opponents
 Designing a Learning System

1.2 Designing a Learning System

– Choosing the training experience:
• Direct (teacher)
• Indirect (credit assignment)
• Distribution of examples
 Designing a Learning System (contd..)

– Choosing the target function

• Legal moves are known a priori, but the best search strategy is not known
• Target function: ChooseMove B  M
B: legal board states M: optimal legal move

• Alternatively: Real function V : B  

Learning task: Discover an operational description of the ideal target function V
(function approximation)
 Designing a learning system
– Choosing a Representation for the Target Function

V(b) = w0 + w1 X1 +...+ w6 X6

X1,2 : Number of black/red pieces on the board

X3,4 : Number of black/red kings on the board

X5,6 : Number of black/red pieces threatened (can be captured on red/black next turn)
 Designing a Learning System (contd..)

– Choosing a Function Approximation Algorithm

• Training examples (b,Vtrain(b))
• Rule for estimating training values:
Vtrain(b)  V[Successor(b)]

– Adjusting the Weights

E   examples
V train ( b )  Vˆ ( b )  2
 1. Introduction
– Design Choices
 1. Introduction

1.3 Some Issues in Machine Learning

– What algorithms can approximate functions well (and when)?
– How does number of training examples influence accuracy?
– How does complexity of hypothesis representation impact it?
– How does noisy data influence accuracy?
– What clues can we get from biological learning systems?
 Recap – Summary – What you have learnt

• The evolution of machine learning over time

• Real-time applications of machine learning
• Defining a machine learning problem as a well posed
learning problem

Wednesday, July 14, 2021 Department of Biotechnology, GIT Course Code and Course Title: 25
 References
• Machine Learning, Tom Mitchell (McGraw Hill, 1997)

Wednesday, July 14, 2021 Department of Biotechnology, GIT Course Code and Course Title: 26
 THANK YOU

Wednesday, July 14, 2021 Department of Biotechnology, GIT Course Code and Course Title: 27