Artificial Intelligence: Definition and Applications

25/06/2022
Welcome! !‫أهﻼ وسهﻼ‬
Artificial Intelligence:
Definition and Applications
• Pr. Hamid BENTARZI

• Laboratory Signals and Systems (LSS)
• IGEE, UMBB University , Boumerdes
• E-mail: h.bentarzi@univ-boumerdes.dz
• ALGERIA
Outline:
1. Introduction
2. Classification of AI Techniques
3. Artificial Intelligence Applications
4. Artificial Intelligence Drawbacks
5. Conclusions
1
25/06/2022
1. Introduction
What is Artificial Intelligence?
 Artificial Intelligence (AI) is the ability of a machine

controlled by a computer to do tasks that are
usually done by humans.
 It is the branch of computer sciences that emphasizes
the development of intelligence machines, thinking and
working like humans. For example, speech
recognition, problem-solving, learning and planning
 John McCarthy is the father of AI.
Artificial Intelligence
(AI) Techniques
Symbolic Knowledge Computational

Representation Knowledge
Representation
Exact Approximate Artificial

Reasoning Reasoning Neural
Network
(ANN)
Expert System Fuzzy Logic

(ES) (FL)
4
2
25/06/2022
2.1 Expert System
Definition: Expert System is a computer program that

uses knowledge and inferences procedures to solve
problems that are ordinarily solved through human
expertise.
2.1 Expert System
Characteristics of Expert Systems

Adavantages Drawbacks
• o Lack of information
o Brittleness (noise)
o Expertise-based shortcomings
o Expert-based shortcomings
3
25/06/2022
2.1 Expert System

What is rules in expert systems?
Rule-based expert systems are expert systems in which the

knowledge is represented by production rules. A production rule, or
simply a rule, consists of an IF part (a condition or premise) and a
THEN part (an action or conclusion). IF condition THEN action
(conclusion).
A classic example of a rule-based system is the domain-specific

expert system that uses rules to make deductions or choices. For
example, an expert system might help a doctor choose the correct
diagnosis based on a cluster of symptoms
2.1 Expert System
Any rule consists of two parts: the IF part,

called the antecedent (premise or condition)
and the THEN part called the consequent
(conclusion or action).
The basic syntax of a rule is:

IF < antecedent >
THEN < consequent>
4
25/06/2022
2.1 Expert System
2.1 Expert System
Structure of Rule-Based Expert System
Knowledge Data
Base Inference Engine Base
(Rules) (facts)
Explanation
Facility
User Interface
10
5
25/06/2022
2.1 Expert System
1- The knowledge base contains the domain knowledge useful for

problem solving. In a rule-based expert system, the knowledge is
represented as a set of rules. Each rule specifies a relation,
recommendation, directive, strategy or heuristic and has the IF
(condition) THEN (action) structure. When the condition part of a
rule is satisfied, the rule is said to fire and the action part is executed.
2- The database includes a set of facts used to match against the IF

(condition) parts of rules stored in the knowledge base.
3- The inference engine carries out the reasoning whereby the expert
system reaches a solution. It links the rules given in the knowledge
base with the facts provided in the database.
11
2.1 Expert System
4- The explanation facilities enable the user to ask the expert system
how a particular conclusion is reached and why a specific fact is
needed. An expert system must be able to explain its reasoning and
justify its advice, analysis or conclusion.
5- The user interface is the means of communication between a user

seeking a solution to the problem and an expert system. The
communication should be as meaningful and friendly as possible.
12
6
25/06/2022
2.1 Expert System
Classification - identify an object based on stated characteristics.
Diagnosis Systems - infer malfunction or disease from observable data.
Monitoring - compare data from a continually observed system to

prescribe behaviour.
Process Control - control a physical process based on monitoring.
Design - configure a system according to specifications.

Scheduling & Planning - develop or modify a plan of action.
Generation of Options - generate alternative solutions to a problem.

13
2. Classification of AI Techniques : 2.2 Fuzzy Logic
What is Fuzzy logic ?

Fuzzy logic is an approach to computing based on "degrees of
truth" rather than the usual "true or false" (1 or 0) Boolean logic
on which the modern computer is based. The idea of fuzzy logic
was first advanced by Lotfi Zadeh of the University of California at
Berkeley in the 1960s.
14
7
25/06/2022
2.2 Fuzzy Logic

Main Components of Fuzzy Logic Reasoning
Output
Input Decision
variables
Fuzzification Fuzzy Defuzzification
Inference
Rules Base
15
2.2 Fuzzy Logic

Its Architecture contains four parts :
1) FUZZIFICATION: It is used to convert inputs i.e. crisp numbers into fuzzy

sets. Crisp inputs are basically the exact inputs measured by sensors and passed
into the control system for processing, such as temperature, pressure, rpm’s, etc.
The process of converting specific input values into degree of membership of
fuzzy sets via membership functions.
2) RULE BASE: It contains the set of rules and the IF-THEN conditions
provided by the experts to govern the decision-making system, on the basis of
linguistic information. Recent developments in fuzzy theory offer several
effective methods for the design and tuning of fuzzy controllers. Most of these
developments reduce the number of fuzzy rules.
16
8
25/06/2022
2.2 Fuzzy Logic

Its Architecture contains four parts :
3) INFERENCE ENGINE: It determines the matching degree of the current

fuzzy input with respect to each rule and decides which rules are to be activated
according to the input field. Next, the activated rules are combined to form the
control actions.
The method of obtaining the final fuzzy conclusion, according to the degree of
membership of input variables to fuzzy sets and the detailed fuzzy rules. There
are many ways about fuzzy inference. Among the methods, the Mamdani
inference method based on Max–Min compositional rule is usually used.
4) DEFUZZIFICATION: It is used to convert the fuzzy sets obtained by the

inference engine into a crisp value. There are several defuzzification methods
available and the best-suited one is used with a specific expert system to reduce
the error.
17
2.2 Fuzzy Logic

Main Components of Fuzzy Logic Reasoning
Inference methods:
Max-Min composition,
Max-Average comp., ..
Output
Input Decision
variables
Fuzzification Fuzzy Defuzzification
Inference
Membership Fuzzy If-Then Rules Defuzzification

functions methods:
If X1 is BIG and X2 is SMALL
Center of area
Then Y is ON,
Center of sums
If X1 is BIG and X2 is BIG
Mean of Maxima,..
Then Y is OFF.
.. 18
9
25/06/2022
2.2 Fuzzy Logic
What is the difference between fuzzy set and crisp set?

Crisp (logic) set defines the value is either 0 or 1.
Fuzzy set defines the value between 0 and 1 including both 0 and 1. It is also
called a classical set. It specifies the degree to which something is true.
A Fuzzy Set is any set that allows its members to have different degree of
membership, called membership function, having interval [0,1].
Membership function :
Membership functions allow us to graphically represent a fuzzy set. The x axis
represents the universe of discourse, whereas the y axis represents the degrees of
membership in the [0,1] interval. Simple functions are used to build membership
functions (Fuzzifiers) such as :
Singleton fuzzifier
Gaussian fuzzifier
Trapezoidal or triangular fuzzifier (it is considered as good Approximators for any
continuous piecewise linear function).
Membership functions characterize fuzziness (i.e., all the information in fuzzy set),
whether the elements in fuzzy sets are discrete or continuous.
Membership functions can be defined as a technique to solve practical problems 19
by experience rather than knowledge.
2.2 Fuzzy Logic
Advantages Disadvantages
•This system can work with any type of •Many researchers proposed different
inputs whether it is imprecise, distorted or ways to solve a given problem through
noisy input information. fuzzy logic which leads to ambiguity.
•The construction of Fuzzy Logic
Systems is easy and understandable. •There is no systematic approach to solve
•Fuzzy logic comes with mathematical a given problem through fuzzy logic.
concepts of set theory and the reasoning
of that is quite simple. •Proof of its characteristics is difficult or
•It provides a very efficient solution to impossible in most cases because every
complex problems in all fields of life as it time we do not get a mathematical
resembles human reasoning and decision- description of our approach.
making.
•The algorithms can be described with •As fuzzy logic works on precise as well
little data, so little memory is required. as imprecise data so most of the time
20
accuracy is compromised.
10
25/06/2022
2.2 Fuzzy Logic

Applications
It is used in the aerospace field for altitude control of spacecraft and
satellites.
It has been used in the automotive system for speed control, traffic
control.
It is used for decision-making support systems and personal evaluation
in the large company business.
It has application in the chemical industry for controlling the pH,
drying, chemical distillation process.
Fuzzy logic is used in Natural language processing and various
intensive applications in Artificial Intelligence.
Fuzzy logic is extensively used in modern control systems such as
expert systems.
Fuzzy Logic is used with Neural Networks as it mimics how a person
would make decisions, only much faster. It is done by Aggregation of
data and changing it into more meaningful data by forming partial 21 truths
as Fuzzy sets.
2.3 Deep Learning and Machine Learning
Since deep learning and machine

learning tend to be used
interchangeably.
As mentioned here, both deep
learning and machine learning
are sub-fields of artificial
intelligence, and deep learning is
actually a sub-field of machine
learning.
22
11
25/06/2022
2.3 Deep Learning and Machine Learning
Deep Learning VS Machine Learning
Deep Learning Machine Learning

(Non- supervised learning) (supervised learning)
o It automates much of the o It is more dependent on

feature extraction piece of the human intervention to learn,
process, o Human experts determine the
o It eliminats some of the hierarchy of features to
manual human intervention understand the differences
required, and between data inputs,
o It is enabling the use of larger o It is usually requiring more
data sets. structured data to learn.
23
2.4 Artificial Neural Networks
24
12
25/06/2022
Artificial Neural Network (ANN) is a system with inputs and

outputs and it was inspired from the biological field and
contains a set of neurons connected together and organized in
many layers in order to solve problems that cannot be solved
by linear computing. Each layer contains a certain number of
neurons and the input of each layer is connected with the
previous layer neurons where the input layer is connected with
the excitation signal. Three types of layers feed forward neural
networks architecture.
25
Three types of layer feed forward neural networks architecture
26
13
25/06/2022
The output of the neuron Y can be calculated by :
27
Rectified Linear Unit (ReLU), sigmoid functions (sigmoid) and Hyperbolic Tangent
(TanH) are used in many applications.
28
14
25/06/2022
Characteristics of Artificial Neural Networks (ANN)
Advantages Drawbacks
• Powerful pattern classification. o Network design using
• Optimization capabilities. trial & error (no. of
• Fast response.
layers, no. of neurons
in hidden layer,
• Fault tolerant (noise).
learning rate, etc.
• Excellent generalization.
o Generation of large
• Trend prediction. training set.
• Good reliability.
29
Deep learning is actually comprised of neural networks. “Deep” in deep

learning refers to a neural network comprised of more than three layers—
which would be inclusive of the inputs and the output—can be considered
a deep learning algorithm. 30
15
25/06/2022
Deep Recurrent Neural Networks (DRNN)

DRNN is a class of deep ANNs. It is a developed and an advanced
version of ANNs. The DRNNs make the previous outputs as new inputs
for the next stage where there are the hidden states.
Architecture of one DRNN cell 31
DRNNs are composed from a number of cells and each cell is called DRNNs cell. The
input of a DRNNs cell is 𝑥𝑡 at timestamp 𝑡, the hidden state is 𝑎𝑡 at timestamp 𝑡 and the
output is 𝑦𝑡 at timestamp 𝑡. The hidden state output 𝑎𝑡−1 of the previous cell is used as an
input besides the input 𝑥𝑡 for the DRNN cell to provide a hidden state output 𝑎𝑡 and an
output 𝑦𝑡.
Architecture of a series of DRNNs cells

32
16
25/06/2022
One-to-One DRNN
It has one input 𝑥 and one output 𝑦 and it is used as conventional ANNs.
33
One-to-Many DRNN
It has one input 𝑥 and many outputs 𝑦 and it has been used especially in music
generation.
34
17
25/06/2022
Many-to-One DRNN
It has many inputs 𝑥 and one output 𝑦 and it has been used in SiS and Protective
relay.
35
Many-to-Many DRNN
It has many inputs 𝑥 and many outputs 𝑦 and it has been used in
classification and recognition.
36
18
25/06/2022
Advantages Disadvantages
o It can process an input of any o The computation in DRNNs is
length , slow ,
o It can handle a variable-length o DRNNs are trained using
sequence input , Stochastic Gradient Descent
o the accuracy of the DRNN Method (SGDM) for
model is not significantly optimization, the great number
affected by the increase in the of hidden layers will cause
size of the inputs. gradient exploding and
o DRNNs computation take into vanishing problems due to the
consideration the saved and derivative computations,
historic information and the o DRNNs cannot capture long-
weights are updated across time. term dependencies. 37
Long-Short Term Memory (LSTM)

Long-Short Term Memory (LSTM), a type of DRNNs, is proposed to
model the long-term dependencies and to solve the exploding and
vanishing problems.
38
19
25/06/2022
 LSTM is based sequence model that can learn and extract

features directly from inputs in time series domain.
 It can be used in a vast range of applications where inputs
are in time series sequence.
 It can be used for fault diagnosis where the measured data
performed in time series sequences.
LSTM is a powerful machine learning tool that has a

remarkable advantage compared to classical neural network
structures, since it captures all information taking into
consideration both features and time. 39
LSTM cell has two outputs: (1) the memory cell state 𝐶𝑡 and
(2)the hidden state ℎ𝑡.
Where,
40
20
25/06/2022
41

Application # 1
The detection and diagnosis of faults in technical systems are of great
practical significance and paramount importance for the safe operation of
the plant.
The early detection of fault can help avoid system shutdown, breakdown
and even catastrophe involving human fatalities and material damage.
Since the operator cannot monitor all variables simultaneously, an

automated approach is needed for the real time monitoring and diagnosis
of the system.
Application uses the artificial neural network based model
(Long-Short Term Memory (LSTM) )
for the fault detection of Pneumatic valve in cooler water spray system in
the industry. The network is developed to detect a totally nineteen faults.
42
21
25/06/2022

Application # 1
43

Application # 2
The application of data-driven models for failure rate estimation.
More specifically, the objectives are to:
1) demonstrate how data-driven methods, i.e. PCA can be used to identify
significant influencing factors for the specific failures of SISs, and
2) propose a framework for predicting the failure rates based on the identified
factors.
“Operational data-driven prediction for failure rates of equipment in

safety instrumented systems: A case study from the oil and gas industry” 44
22
25/06/2022

Application # 3
Fuzzy logic deals with uncertainty and imprecision, and is an efficient
tool for problems where knowledge uncertainty may occur. Such
situations arise frequently in a quantitative fault and event tree analysis
in safety and risk assessment of different processes.
Application # 4
Cyber security is not a single problem, but rather it is a group of highly
different problems involving different sets of threats. An Advance Cyber
Security System (ACSS) using fuzzy logic is a system that consists of a
rule depository and a mechanism for accessing and running the rules.
The depository is usually constructed with a collection of related rule
sets. Fuzzy optimization deals with finding the values of input
parameters of a complex simulated system which results in desired
output. Fuzzy logic controller is used to execute fuzzy logic inference
rules from a fuzzy rule base in determining the congestion parameters,
45
getting the warning information and the appropriate action.

Application # 5
AI associated with Industrial Internet of Things (IIoT) employs

networked sensors to collect critical data and uses cloud software to turn
this data into valuable insights about the efficiency and the safety of the
petroleum industry operations.
Improved safety. IIoT with AI help to ensure a safer workplace. Paired

with wearable devices, IIoT allows monitoring workers’ health state and
risky activities that can lead to injuries. Along with ensuring workers’
safety, IIoT with AI address safety problems in potentially hazardous
environments. For instance, in petroleum industry, IIoT with AI are
applied to monitor gas leakages as it travels through the pipe network.
46
23
25/06/2022

Application # 5
IIoT with AI
47

Application # 5
IIoT with AI
48
24
25/06/2022

Application # 5
IIoT with AI
He further goes on to assert that another practical way of improving

safety is through wearable sensors, which can be worn over a jacket, on the
wrist or on the helmet. These sensors can be streaming data & monitoring
activities that are going on in a situation in the factory that might be hazardous. If
one sees occurrences through Machine Learning and data analytics that are
not routine, that can also create a warning before a serious accident can
happen. Just like in automated vehicles or self-driven cars, these kinds of
technologies can also be applied inside a factory where multiple forklift
machines can actually talk to each other through sensors via advanced
commmunication technologies such as 5G, WiFi, LoRa. They can avoid
collisions and hitting objects inside the plant.
49

Application # 5
IIoT with AI
50
25
25/06/2022

Cyber attacks due to the reduction of security risk level
TRITON: The First ICS Cyber Attack on Safety Instrument Systems
51

In normal machine learning (left), the learning algorithm extracts
patterns from a dataset, and the “learned” knowledge is stored in the machine
learning model—the brain of the system. In a poisoning attack (right), the
attacker changes the training data to poison the learned model.
52
26
25/06/2022

In regular use (top), the AI system takes a valid input, processes it with the
model (brain), and returns an output.
However, in an input attack (bottom), the input to the AI system is altered
with an attack pattern, causing the AI system to return an incorrect output.
53
Conclusions
Benefit of AI :Improved monitoring. AI's capacity to process data in
real time means organizations can implement near-instantaneous
monitoring; that can enhance both industrial control as well as
industrial protection and safety.
Risk of AI
AI errors: While AI can eliminate human error,
However, problematic data, poor (poisoning) training data or mistakes
in the algorithms can lead to AI errors. And those errors can be
dangerously compounded because of the large volume of
transactions AI systems typically process.
"Humans might make 30 mistakes in a day, but a bot handling
millions of transactions a day magnifies any error,“
Security risk level reduction : Cyber attacks
54
27
25/06/2022
Conclusions
The use of AI can bring the benefits to the industry such as
improvement in the controller performance, protection,
safety, and energy efficiency, while meeting industrial
expectations of security and reliability.
55
http://igee.univ-boumerdes.dz/Laboratoires.html
E-mail: h.bentarzi@univ-boumerdes.dz
56
28

Artificial Intelligence: Definition and Applications

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Artificial Intelligence: Definition and Applications

Uploaded by

Copyright:

Available Formats

25/06/2022

Welcome! !‫أهﻼ وسهﻼ‬

• Pr. Hamid BENTARZI

 Artificial Intelligence (AI) is the ability of a machine

Symbolic Knowledge Computational

Exact Approximate Artificial

Expert System Fuzzy Logic

Definition: Expert System is a computer program that

Characteristics of Expert Systems

2.1 Expert System

Rule-based expert systems are expert systems in which the

A classic example of a rule-based system is the domain-specific

2.1 Expert System

Any rule consists of two parts: the IF part,

The basic syntax of a rule is:

2.1 Expert System

Structure of Rule-Based Expert System

2.1 Expert System

1- The knowledge base contains the domain knowledge useful for

2- The database includes a set of facts used to match against the IF

2.1 Expert System

5- The user interface is the means of communication between a user

2.1 Expert System

Classification - identify an object based on stated characteristics.

Diagnosis Systems - infer malfunction or disease from observable data.

Monitoring - compare data from a continually observed system to

Process Control - control a physical process based on monitoring.

Design - configure a system according to specifications.

Generation of Options - generate alternative solutions to a problem.

2. Classification of AI Techniques : 2.2 Fuzzy Logic

What is Fuzzy logic ?

2.2 Fuzzy Logic

2.2 Fuzzy Logic

1) FUZZIFICATION: It is used to convert inputs i.e. crisp numbers into fuzzy

2.2 Fuzzy Logic

3) INFERENCE ENGINE: It determines the matching degree of the current

4) DEFUZZIFICATION: It is used to convert the fuzzy sets obtained by the

2.2 Fuzzy Logic

Membership Fuzzy If-Then Rules Defuzzification

2.2 Fuzzy Logic

What is the difference between fuzzy set and crisp set?

2.2 Fuzzy Logic

2.2 Fuzzy Logic

Since deep learning and machine

Deep Learning Machine Learning

o It automates much of the o It is more dependent on

Artificial Neural Network (ANN) is a system with inputs and

Three types of layer feed forward neural networks architecture

The output of the neuron Y can be calculated by :

Characteristics of Artificial Neural Networks (ANN)

Deep learning is actually comprised of neural networks. “Deep” in deep

Deep Recurrent Neural Networks (DRNN)

Architecture of one DRNN cell 31

Architecture of a series of DRNNs cells

2.4 Artificial Neural Networks

Long-Short Term Memory (LSTM)

2.4 Artificial Neural Networks

Long-Short Term Memory (LSTM)

 LSTM is based sequence model that can learn and extract

LSTM is a powerful machine learning tool that has a

2.4 Artificial Neural Networks

Long-Short Term Memory (LSTM)

3. Artificial Intelligence Applications

3. Artificial Intelligence Applications