Basics and introduction to Artificial Intelligence

2. Introduction
• AI has always intrigued people.
• For example, conversing with a computer using natural
language has always been a dream for humanity.
• Science fiction has long anticipated the rise of machine
intelligence. Remember Terminator, AI, Her, Space
Odyssey, Matrix and Ex machina and so on and so forth
• Today, a new generation of
self-learning computers is reshaping
every aspect of our lives.

3. Sequence
• Introduction
• Definition
• History / evolution
• Branches (Sub field) of AI
• Classification/ types
– Narrow AI
– Artificial General Intelligence
– Artificial Super Intelligence
• Applications, examples
• Languages : Python, Java, Lisp

4. Artificial is something
that is not real and which is kind of
fake because it is simulated.
Perception, reasoning, learning,
communicating, acting in complex
environments, self-awareness, adapting,
creativity and problem solving

5. Definition
• AI is intelligence exhibited by machines
• Branch of computer science which deals with
creating computers or machines as intelligent as
human beings
• Artificial intelligence is a wide field of new types
of computers that are trained rather than
programmed
• AI is concerned with intelligent behaviour in
artifacts. (Artificial Intelligence – A new synthesis
by Nilsson, 1998)

6. Definition

7. Definition
• Automatic
– Execution of precise, repetitious actions or
sequence in controlled or well understood
environment
– Pre programmed
• Autonomous
– Generation and execution of actions to meet a
goal or carry out a mission
– Adaptive
– Self decision making

8. Definition
• Smart device
– an electronic device, generally connected to other
devices or networks via different wireless protocols
that can operate to some extent interactively and
autonomously.
– Term can also refer to a device that exhibits some
properties of ubiquitous computing, including—
although not necessarily—artificial intelligence.
• AI
– Designed to constantly seek patterns (like humans),
learn from experience (like humans) and self-select
the appropriate responses in situations based on that
(like humans)

9. History
1943:Foundation of Neural Network
established by Warren McCulloch and
Walter Pitts
Drawing parallels between the brain
and computing machines

10. History
In 1950, Alan Turing introduces the test-
A way of testing a machine’s intelligence
Computers those indistinguishable from human
beings
Turing predicted intelligent computing in 50
years

11. History
• The term Artificial Intelligence was
first coined by John McCarthy in
Dartmouth conference devoted to the
topic in 1956

12. History
• 1958 In MIT lab,
McCarthy defined the
high level language
Lisp, a dominant AI
programming language
• 1961 Newell and
Simon's presented
the GPS, a program
to imitate human
problem-solving
protocols

13. History
1965 ELIZA a natural language program
created by Joseph Weizenbaum at MIT
ELIZA handles dialogue on any topic,
similar in concept to todays chatbot

14. History
Dendral is considered the first expert
system
Developed at Stanford University by Edward
Feigenbaum, Bruce G. Buchanan, Joshua
Lederberg
It automated the decision-making process
and problem-solving behavior of organic
chemists.
Used to identify the structure of chemical
compounds

15. History
1970 MYCIN expert system in medical
domain
Developed at Stanford University by Edward
Shortliffe
Identify bacteria causing severe
infections, and to recommend antibiotics,
diagnosis of blood clotting diseases
It was written in Lisp

16. History
Edward Feigenbaum, "father of expert
systems“
1980-1985 proliferation of expert
systems

17. History
• AI winter 1985-95
• Emergence of Intelligent agents
• 1997 Computer Deep Blue beats world chess champion
Gary Kasparov
• Availability of large data sets, distributive computing,
Explosion of Internet
• Processing power, Proliferation of GPU
• 2002 Irobot Roomba autonomous vacuum cleaner
• ASIMO humanoid robot created by Honda in 2000
• 2009 Google build s first self driving car

18. History
2011 : IBM,s Watson defeats champions of US
game show Jeopardy
2011-14 : Personal assistants like Siri,
Google Now, Cortana use speech recognition
to answer questions and perform simple tasks
2016 : ALphaGO beats professional Go
player Lee Sedol
Artificial intelligence systems take on more
tasks and solve more problems

19. Types of Artificial Intelligence
Generally, A.I. falls within three categories —
• Artificial Narrow Intelligence
(ANI)
• Artificial General Intelligence
(AGI)
• Artificial Super Intelligence
(ASI)

20. Narrow Artificial Intelligence
Developed and trained for a particular, single task and
works within a limited context
Narrow AI can identify pattern and correlations from data
more efficiently than humans.
• Eg. Sales predictions, purchase suggestions and weather
forecast etc.
But it cannot expand and take tasks beyond its field
• Eg. for example, AI engine which transcripts image
recognition cannot perform sales recommendations.
Several narrow A.I.s can be strung together to offer a more
comprehensive service:
• Eg. Alexa, Google Assistant, Siri, and Cortana are great
examples

21. Artificial General Intelligence
• Strong AI
• AI system with generalized cognitive abilities which
find solutions to the unfamiliar task it comes across.
• ability to understand context and make judgments
based on it
• Over time, it learns from experience, is able to make
decisions even in times of uncertainty or with no prior
available data, use reason, and be creative.
• Intellectually, these computers operate much like the
human brain.
• So far we’ve not been able to do it, although most
believe we might be able to do so sometime in future

22. Artificial Super Intelligence:
• Artificial Super Intelligence (ASI) refers to the position
where computer/machines will surpass humans and
machines would be able to mimic human thoughts.
• ASI refers to a situation where the cognitive ability of
machines will be superior to humans.
• A.I. robots would be able to think for themselves, attain
consciousness, and operate without any human
involvement, perhaps at the direction of another A.I

23. Artificial Super Intelligence:
• There is still no machine that can process the depth of
knowledge and cognitive ability as that of a fully
developed human.
• ASI had two school of thoughts,
– On one side great scientist like Stephen Hawking saw
the full development of AI as a danger to humanity
– Demis Hassabis, Co-Founder & CEO
of DeepMind believes that smarter the AI becomes
better the world would be and a helping hand to
mankind.

24. Most Popular AI Languages
Python
Lisp
Prolog
Java
C++

25. Types of AI
• Expert Systems
• Machine Learning
– Artificial Neural Network (Deep Learning)
• Natural Language Processing
– Speech Recognition
• Evolutionary computation incl Genetic Algorithm
• Fuzzy Systems
• Robotics
• Computer Vision
– Image Recognition

26. Expert Systems
• An expert system is a computer system that
emulates, or acts in all respect, with the
decision making capabilities of a human
expert
• Knowledge Based System
• Expert to have extensive knowledge

27. Machine Learning
• Machine Learning is the field of study that gives
computers the capability to learn without being
explicitly programmed.
Arthur Lee Samuel
• More similar to humans: The ability to learn.
• Machine learning is about predicting the future
based on the past.
• The process starts with feeding good quality data
and then training machines(computers) by building
machine learning models using the data and
different algorithms.

28. Machine Learning

29. Machine Learning
• Supervised learning: Supervised learning is the learning of
the model where we have input variable ( say, x) and
corresponding output variable (say, Y) i.e. training data set
• Aim is to approximate the mapping function so well that when
there is a new input data (x) then the corresponding output
variable can be predicted.
• Unsupervised Learning: Unsupervised learning is where
only the input data (say, X) is present and no corresponding
output variable is there.
• Aim of Unsupervised learning is to model the distribution in
the data in order to learn more about the data.

30. • Reinforcement Learning is a type of Machine Learning
in which a learning algorithm is trained not on preset
data but rather based on a feedback system.
• There’s no answer key, but based on trial and error
reward or penalty is awarded
• Resultantly computer improves itself over a period of
time by many training data set
• Reinforcement learning is all about making decisions
sequentially. In simple words we can say that the output
depends on the state of the current input and the next
input depends on the output of the previous input

32. Artificial neural networks
• ..a computing system made up of a number of
simple, highly interconnected processing
elements, which process information by their
dynamic state response to external inputs.”
Dr. Robert Hecht-Nielsen

33. Biological Neuron

34. Artificial neurons
one possible model
Inputs
Output
w2
w1
w3
wn
wn-1
.
.
.
x1
x2
x3
…
xn-1
xn
y
)
(
;
1
z
H
y
x
w
z
n
i
i
i 
 


35. Artificial neural networks
• ANNs are composed of multiple nodes, which imitate
biological neurons of human brain. The neurons are
connected by links and they interact with each other.
• The nodes can take input data and perform simple
operations on the data. The result of these operations
is passed to other neurons. The output at each node is
called its activation or node value.
• Each link is associated with weight. ANNs are capable
of learning, which takes place by altering weight
values.

36. Artificial neural networks
Inputs
Output
An artificial neural network is composed of many artificial neurons that are
linked together according to a specific network architecture. The objective of
the neural network is to transform the inputs into meaningful outputs.

37. Fuzzy Systems
• The term fuzzy logic was first used in 1965 by Lotfi
Zadeh, a professor of UC Berkeley
• Fuzzy mean things which are not very clear or vague
• Concepts which do not have sharply defined boundaries
– for example, many, tall, much larger than, young, etc.
• Allow the computer to determine the distinctions among
data which is neither true nor false. Something similar to
the process of human reasoning. Like Little dark, Some
brightness,
• Fuzziness is that it allows a gradual and continuous
transition, say, from 0 to 1

38. Fuzzy Systems
• Uncertainty is found to arise from ignorance,
from chance and randomness, due to lack of
knowledge existing in our natural language.
• Fuzzy systems are suitable for uncertain or
approximate reasoning
– allows decision making with estimated values under
incomplete information
• It provides a technique to deal with imprecision
and information granularity.

39. • Fuzzy control of various physical or chemical
characteristics such as temperature, electric current, flow
of liquid/gas, motion of machines, etc.
• Fuzzy logic is not always accurate, so The results are
perceived based on assumption, so it may not be widely
accepted.

41. Genetic Algorithm
• Genetic Algorithm (GA) is a optimization
technique
• Optimization refers to finding the values of inputs
in such a way that we get the “best” output
values.
• It is frequently used to find optimal or near-
optimal solutions to difficult problems which
otherwise would take a lifetime to solve.
• GAs were developed by John Holland and his
students and colleagues at the University of
Michigan

42. Genetics Algorithm
• In GAs, we have a pool or a population of possible
solutions to the given problem. These solutions then
undergo recombination and mutation (like in natural
genetics), producing new children, and the process is
repeated over various generations. Each individual (or
candidate solution) is assigned a fitness value (based
on its objective function value) and the fitter
individuals are given a higher chance to mate and yield
more “fitter” individuals.
• In this way we keep “evolving” better individuals or
solutions over generations, till we reach a stopping
criterion.

43. Genetic Algorithm
• Advantages
– Is faster and more efficient as compared to the traditional methods.
– Has very good parallel capabilities.
– Provides a list of “good” solutions and not just a single solution.
– Always gets an answer to the problem, which gets better over the
time.
– Useful when the search space is very large and there are a large
number of parameters involved.
• Limitations of GAs
– Being stochastic, there are no guarantees on the optimality or the
quality of the solution.
– If not implemented properly, the GA may not converge to the
optimal solution.

44. Computer Vision
• The goal of computer vision is to develop
algorithms that allow computer to “see”.
• Also called
• Image Understanding
• Image Analysis
• Machine Vision

45. Autonomous Vehicle
Flakey, SRI

46. Applications: Robotics

47. Applications: Surveillance

48. Hardware
• GPU
• ASICs
• FPGA
• AI Accelerators

49. Hardware
• HPE Apollo 6500 Gen10 System
• Dell PowerEdge R740xd
• IBM Power System™ Accelerated Compute
Server (AC922)

50. Joint Artificial Intelligence Center
(JAIC)
• Joint Artificial Intelligence Center (JAIC) is the
Department of Defense’s (DoD) Artificial
Intelligence (AI) Center of Excellence that
provides a critical mass of expertise to help
the Department harness the game-changing
power of AI.

51. Joint Artificial Intelligence Center
(JAIC)
• Accelerating the delivery and adoption of AI
• Scaling the impact of AI across the Department
• Defend U.S. critical infrastructure from malicious cyber
activity that alone, or as part of a campaign, could
cause a significant cyber incident
• Establishing a common foundation that enables
decentralized execution and experimentation
• Evolving partnerships with industry, academia, allies
and partners
• Cultivating a leading AI workforce
• Leading in military AI ethics and safety

52. Air Force Cognitive Engine (ACE)
• AI software eco system
• Air Force researchers working with artificial
intelligence code may soon have a platform that
gives them secure access to educated end-users
and outside developers, algorithms, mission data
and computational hardware.
• ACE is in developmental beta phase, which will
help shape the program's architecture going
forward. Production version 1.0 is expected to be
released summer 2020.

53. Air Combat Evolution (ACE)
• Automate air-to-air combat as part of its Air Combat
Evolution (ACE) programme.
• Promoting human-machine collaborative dogfighting
• Aerial dogfighting over to AI as it hopes the technology
will be able to handle a high-end fight, elevating the
pilot’s role to cockpit-based mission commander.
• Artificial Intelligence (AI) to develop autonomous air-to-
air combat capabilities
• Known as ‘mosaic warfare’, this approach involves
linking manned aircraft together with inexpensive
unmanned systems to fight the combat.

54. • AI technologies will be trained in aerial dogfighting in a
manner similar to how new pilots are trained.
• The training will include basic fighter maneuvers in
simple, one-on-one scenarios. If pilots are satisfied with
the reliability of the AI algorithms in handling bounded,
transparent and predictable behaviours, the training will
proceed to more complex aerial engagement scenarios
• Concept for the Skyborg UAV released by the USAF. DARPA is
exploring using such ‘loyal wingmen’ in the air-to-air combat role,
with control coming from manned aircraft.

55. USAF
• Project Maven
• ALPHA: A high-fidelity air combat simulator by
Psibernetix
• MIT and the U.S. Air Force have signed an agreement to
launch a new program designed to make fundamental
advances in artificial intelligence that could improve Air
Force operations while also addressing broader societal
needs.
– The effort, known as the MIT-Air Force AI Accelerator, will
leverage the expertise and resources of MIT and the Air Force to
conduct fundamental research directed at enabling rapid
prototyping, scaling, and application of AI algorithms and
systems.

56. USA
• Autonomous Weapons and Weapons
Targeting
• Surveillance
• Cybersecurity
• Homeland Security
• Logistics
• Autonomous Vehicles

57. China
• Autonomous Vehicles and Drones
• “New Generation Artificial Intelligence
Development Plan” This policy outlines
China’s strategy to build a domestic AI
industry worth nearly US$150 billion in the
next few years and to become the leading AI
power by 2030