AI

vs
Machine Learning
vs
Deep Learning
Objectives

1. Define artificial intelligence, machine learning and deep

learning
2. Explain how deep learning solves classical machine learning
limitations
3. Explain key historical developments
4. Differentiate modern AI from prior AI
5. Relate sample applications of AI
AI Breakthroughs

Image classification Machine translation

AI as the New Electricity

“About 100 years ago, electricity

transformed every major industry. AI
has advanced to the point where it
has the power to transform…every
major sector in coming years.”
-Andrew Ng, Stanford University
● Artificial intelligence
● Machine learning
● Deep learning
Artificial Intelligence

“A branch of “Colloquially, the term

computer science ‘artificial intelligence’ is
dealing with the applied when a
“A program that machine
simulation of can sense, reason, mimics ‘cognitive’
intelligent act, and adapt.” functions that humans
behavior in associate with other
computers.” (Intel) human
minds, such as
(Merriam-Webster) ‘learning' and ‘problem
solving’.” (Wikipedia)
Machine Learning

“The study and construction of

programs that are not explicitly
programmed, but learn patterns
as they are exposed to more data
over time.”

(Intel)
Machine Learning

These programs learn from repeatedly seeing data,

rather than being explicitly programmed by humans.
Machine Learning Technology

This example is learning to classify a species from a set of

measurement features
Machine Learning Example

● Suppose you wanted to identify

fraudulent credit card
transactions
● You could define features to be:
○ Transaction time
○ Transaction amount
○ Transaction location
○ Category of purchase
● The algorithm could learn what
feature combinations suggest
unusual activity
Machine Learning Limitations

● Suppose you wanted to

determine if an
image is of a cat or a dog
● What features would you
use?
● This is where Deep
Learning can come in
Dog and cat recognition
Deep Learning

● “Machine learning that

involves using very
complicated models called
“deep neural networks”."
(Intel)

● Models determine best

representation of original
data; in classic machine
learning, humans must do
this
Machine learning vs Deep learning
Deep Learning Example
History of AI

AI has experienced several hype cycles, where it has oscillated

between periods of excitement and disappointment
1950s: Early AI

1950 1956 1957 1959

Alan Turing developed Artificial Intelligence Frank Rosenblatt

the Turing test to was accepted as a invented the Arthur Samuel
test a machines ability field at the Dartmouth perceptron algorithm. published an algorithm
to exhibit intelligent Conference This was the precursor for a checkers program
behavior to modern neural using machine
networks learning
The First “AI Winter”

ALPAC committee evaluated AI The Lighthill report highlights AI’s

techniques for machine failure to live up to promises
translation and determined
there was little yield from the
investment

1969

1966 1973
Marvin Minsky published a The two reports led to cuts in
book on the limitations of the government funding for AI research
Perceptron algorithm which leading to the first “AI Winter.”
slowed research in neural
networks
1980’s AI Boom

1986: The
Were the first “Backpropogation”
Expert Systems - Ran on mainframe widely-used AI algorithm is able to
systems with computers with technology, with train multi-layer
programmed specialized two-thirds of "Fortune perceptrons leading to
rules designed to mimic programming 500" companies using new successes and
human experts languages (e.g. LISP) them at their peak interest in neural
network
research
Another AI Winter (late 1980’s – early 1990s)

● Expert systems’ progress on solving

business problems slowed
● Expert systems began to be melded into
software suites of general business
applications (e.g. SAP, Oracle) that could
run on PCs instead of mainframes
● Neural networks didn’t scale to large
problems
● Interest in AI in business declined
Late 1990’s to early 2000’s: Classical Machine Learning

AI algorithms
Advancements in AI solutions had were integrated The Deep Blue Google search
the SVM successes in into larger chess system engine launched
algorithm led to it speech systems beat world chess using artificial
becoming the recognition, and became champion Garry intelligence
machine learning medical diagnosis, useful Kasparov technology
method of choice robotics, and throughout
many other areas industry
2006: Rise of Deep Learning

● 2006: Geoffrey Hinton publishes a paper on unsupervised

pre-training that allowed deeper neural networks to be trained
● Neural networks are rebranded to deep learning
● 2009: The ImageNet database of human-tagged images is
presented at the CVPR conference
● 2010: Algorithms compete on several visual recognition tasks at the
first ImageNet competition
Deep Learning Breakthroughs (2012 – Present)

Deep learning beats Similar breakthroughs

previous benchmark on the appeared in language
ImageNet competition translation

2012 2014

2013

These have led to

Deep learning is used to advancements in Web
understand “conceptual Search, Document Search,
meaning” of words Document Summarization,
and Machine Translation
Deep Learning Breakthroughs (2012 – Present)
2014
Computer vision algorithm can describe photos

2015
Deep learning platform TensorFlow* is developed

2016
DeepMind* AlphaGo developed by Aja Huang, beats
Go master Lee Se-dol
Modern AI (2012 – Present): Deep Learning Impact

Computer vision Natural language

How Is This Era of AI Different?
Other Modern AI Factors

● Continued expansion of open source AI,

especially in Python*, aiding machine
learning and big data ecosystems
● Leading deep learning libraries open
sourced, allowing further adoption by
industry
● Open sourcing of large datasets of millions
of labeled images, text datasets such as
Wikipedia has also driven breakthroughs
Transformative Changes
Transformative Changes
Transformative Changes
Transformative Changes
Some more
applications
AI Omnipresence In Transportation

Navigation Ride Sharing

AI Omnipresence In Transportation

Audience Content
AI Omnipresence In Transportation

Natural Language Object Detection

Latest Developments: Computer Vision
Application Area: Abandoned Baggage Detection

● We can automatically detect

when baggage has been left
unattended, potentially saving
lives
● This system relies on the
breakthroughs:
○ Cutting edge object
detection
○ Fast hardware on which to
train the model, e.g. Intel
RTX 2070 processors
References:
Alpha Go: https://www.youtube.com/watch?v=8dMFJpEGNLQ
Simplilearn - https://www.youtube.com/watch?v=9dFhZFUkzuQ
Hans-D, W. (2017). Machine Learning, Deep Learning and AI: What’s the Difference
https://en.wikipedia.org/wiki/Artificial_intelligence
https://en.wikipedia.org/wiki/Machine_learning
https://en.wikipedia.org/wiki/Deep_learning
https://machinelearningmastery.com/what-is-deep-learning/
https://www.ibm.com/cloud/learn/deep-learning
https://www.g2.com/articles/history-of-artificial-intelligence
https://www.javatpoint.com/applications-of-machine-learning
https://www.simplilearn.com/tutorials/deep-learning-tutorial/deep-learning-applications