Scribd
Upload
132 views

Introduction To Machine Learning - 2023

Machine learning is a set of methods that can automatically detect patterns in data and use those patterns to predict future data or make decisions. It works by learning from examples rather than being explicitly programmed. There are different types of machine learning including supervised learning (prediction based on labeled examples), unsupervised learning (finding hidden patterns in unlabeled data), and reinforcement learning (learning from rewards/punishments). Machine learning is being applied to many domains like computer vision, natural language processing, and speech recognition by developing algorithms that can learn from large amounts of data.

Uploaded by

cawifi4523
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
132 views

Introduction To Machine Learning - 2023

Machine learning is a set of methods that can automatically detect patterns in data and use those patterns to predict future data or make decisions. It works by learning from examples rather than being explicitly programmed. There are different types of machine learning including supervised learning (prediction based on labeled examples), unsupervised learning (finding hidden patterns in unlabeled data), and reinforcement learning (learning from rewards/punishments). Machine learning is being applied to many domains like computer vision, natural language processing, and speech recognition by developing algorithms that can learn from large amounts of data.

Uploaded by

cawifi4523
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 44

Introduction to Machine Learning

Dr. Srijith P K
Computer Science and Engineering
IIT Hyderabad
• “A breakthrough in machine learning would be worth ten Microsofts” (Bill
Gates, Chairman, Microsoft)
• •“Machine learning is the next Internet” (Tony Tether, Director, DARPA)
• “AI is one of the most important things humanity is working on. It is more
profound than, I don't know, electricity or fire.” - Sundar Pichai (Google CEO)
Big data Era

• Google: 3.5 billion search queries


every day.2. •1.7MB of data is created every second by every person
during 2020.
• Facebook: 350 million photos are
uploaded to Facebook each day. •In the last two years alone, the astonishing 90% of the
Facebook generates 4 petabytes of world’s data has been created.
data every day. •2.5 quintillion bytes of data are produced by humans every
day.
• Every day, 306.4 billion emails are
sent, and 5 million Tweets are made •463 exabytes of data will be generated each day by humans
in Twitter: as of 2025.
•95 million photos and videos are shared every day on
• Astronomy: Satellite data is in
Instagram.
hundreds of PB.
•By the end of 2020, 44 zettabytes will make up the entire
digital universe.
•Every day, 306.4 billion emails are sent, and 5 million
Tweets are made.
https://techjury.net/blog/how-much-data-is-created-every-day/#gref
Why machine learning : Data every where !
“We are drowning in information and
starving for knowledge.” — John Naisbitt.

• This deluge of data calls for automated methods of data analysis, which is what
machine learning provides.
• Defined as a “set of methods that can automatically detect patterns in data, and
then use the uncovered patterns to predict future data, or to perform other kinds of
decision making under uncertainty” - Kevin Murphy (Machine Learning : A
Probabilistic Perspective)
Whats Machine
Learning ?
• Connect the variable voltage supply to both the ends of the
rheostat.Connect the ammeter in series of the
rheostat.Connect the voltmeter in parallel of the
rheostat.Start measuring the voltage and current as you move
the rheostat moving hand from minimum position to the
maximum position in the steps of constant increase in
current.

Estimate
Resistance
Estimating Resistance
is machine learning !
• A currentvoltage characteristic or IV
curve (currentvoltage curve) is a
relationship, typically represented as a
chart or graph, between the electric
current through a circuit, device, or
material, and the corresponding
voltage, or potential difference across
it. In the graph, the voltage is plotted
along the y-axis and the current is
plotted along the x-axis.
V=RI

resistance is 2.5 ohms!


Estimating Resistance
is machine learning !
• A currentvoltage characteristic or IV
curve (currentvoltage curve) is a
relationship, typically represented as a
chart or graph, between the electric
current through a circuit, device, or
material, and the corresponding
voltage, or potential difference across
it. In the graph, the voltage is plotted
along the y-axis and the current is
plotted along the x-axis.
V=RI
Output Input
y=mx
• Algorithms which let machines
What’s machine learning ? learn like humans from
observations !
• To discover the fundamental
principles of learning from data
and use them to develop
algorithms that can learn like
living beings !
• Programming computers to
optimize a performance
criterion using example data or
past experience(Ethem Alpay
din, Machine Learning, 2010)
• How do we create computer
programs that improve with
experience? (Tom Mitchel)
Machine learning
ML based applications

-Machine learning has


become prominent
approach to solve problems
in AI domains like computer
vision, language and speech
processing
ML based
applications
• Machine learning has become
prominent approach to solve problems
in AI domains like computer vision,
language and speech processing

• Early approaches to AI was based on


logic but applications has to face a lot
of uncertain situations and has to
perform well on unseen situations.

• Machine learning focused on


developing algorithms which could
perform well on future unseen data
(generalization performance) which
differentiates it from statistics
Machine learning is interdisciplinary
Overview of
Machine learning

Supervised learning
• Predict an output y when given an input x
• For categorical y : classification.
• For real-valued y : regression.
Unsupervised learning
• Create an internal representation of the input, e.g.
clustering, dimensionality reduction
• This is important in machine learning as getting
labels is often difficult and expensive
Other settings of ML
• Reinforcement learning (learning from “rewards”)
• Semi-supervised learning (combines supervised +
unsupervised)
• Active learning, online learning, Transfer learning,
multi-task learning, Structured prediction
ML in practice
Input and output y a label. Learn a function
Output
Supervised Input

learning
(classification)
Supervised learning (Classification )
Input and output y a real value. Learn a function

Supervised
learning
(Regression) Output

Goal is to learn a function which maps inputs to outputs so


that it will predict well on future data points – Input
Generalization performance
Learn the function which passes through as many points as possible : Minimize the Least Squares Error
Machine learning
algorithms

Method : Linear regression, support vector regression, gaussian


process regression
Training and Testing ML models

ML models should be able to predict well (generalization ability)


Model with minimum error on train data need not be the correct model
Generalization ability
and overfitting
Classification

Regression
Machine learning

• Main references:
• Mitchell, Tom. Machine Learning. New York, NY: McGraw-Hill, 1997
• Alpaydin, Ethem Introduction to Machine Learning. MIT Press, 2014.
• Bishop, Christopher M. Pattern Recognition and Machine Learning. Springer, 2006.
• Murphy, K. P. (2013). Machine learning : a probabilistic perspective. Cambridge, Mass.
MIT Press
• Hastie, T., R. Tibshirani, and J. H. Friedman. The Elements of Statistical Learning: Data Mining,
Inference and Prediction. New York, NY: Springer, 2001

You might also like