Sie sind auf Seite 1von 20

Artificial Intelligence for

Risk Management
Other books by Archie Addo published by Business Expert Press

Artificial Intelligence Design and Solution for Risk and Security


ISBN: 9781951527488
https://www.businessexpertpress.com/books/artificial-intelligence-
design-and-solution-for-risk-and-security/

Artificial Intelligence for Security


ISBN: 9781951527266
https://www.businessexpertpress.com/books/artificial-intelligence-for-
security/
Artificial Intelligence for
Risk Management

Archie Addo, Srini Centhala, and


Muthu Shanmugam
Artificial Intelligence for Risk Management

Copyright © Business Expert Press, LLC, 2020.

Cover image licensed by Ingram Image, StockPhotoSecrets.com

All rights reserved. No part of this publication may be reproduced,


stored in a retrieval system, or transmitted in any form or by any
means—electronic, mechanical, photocopy, recording, or any other
except for brief quotations, not to exceed 400 words, without the prior
permission of the publisher.

As part of the Business Law Collection, this book discusses general princi-
ples of law for the benefit of the public through education only. This book
does not undertake to give individual legal advice. Nothing in this book
should be interpreted as creating an attorney-client relationship with the
author(s). The discussions of legal frameworks and legal issues is not in-
tended to persuade readers to adopt general solutions to general problems,
but rather simply to inform readers about the issues. Readers should not
rely on the contents herein as a substitute for legal counsel. For specific
advice about legal issues facing you, consult with a licensed attorney.

First published in 2020 by


Business Expert Press, LLC
222 East 46th Street, New York, NY 10017
www.businessexpertpress.com

ISBN-13: 978-1-94944-351-6 (paperback)


ISBN-13: 978-1-94944-352-3 (e-book)

Business Expert Press Business Law and Corporate Risk Management


­Collection

Collection ISSN: 2333-6722 (print)


Collection ISSN: 2333-6730 (electronic)

Cover and interior design by Exeter Premedia Services Private Ltd.,


Chennai, India

First edition: 2020

10 9 8 7 6 5 4 3 2 1

Printed in the United States of America.


Abstract
The volcano disaster in Hawaii was uncontrollable (Wall Street Journal
May 2018) and loss of market shares led to a collapse in the economy of
the environment (Pelling et al. 2002; Eddie Guidry et al. 2013). These
kinds of uncertainties require some approach to mitigate the situation.
Experts are required to identify the associated risks early enough to safe-
guard the situation and minimize the impact. Here, risk management
plays a major role in situations of uncertainty. Relevant questions are:
How can this situation be forecasted and taken care of early in sim-
ilar situations? Can these occurrences be captured historically? Can
such patterns of occurrences be identified? Can the data relating to the
occurrences be captured? Can the data be used to predict future occur-
rences? Can the relevance of the data be an important factor? Is secu-
rity important? What happens if the data are manipulated? How are the
data manipulated? How can the data be protected and secured? Secu-
rity takes an important role in the data and is the driving factor in risk
management.
Various types of risk apply to various industries, various business
functions, roles, and responsibilities. This book intends to illustrate top
business cases and use cases that apply to respective industries by sug-
gesting ways to define, analyze, monitor, control, and mitigate risk. The
importance of this approach is to mitigate risk using data and putting
corrective action in place. Because humans cannot quickly analyze huge
amounts of data, such analysis takes a long time. People can use data
science, data analytics, and machine learning (ML) algorithms to speed
the process.
Artificial intelligence (AI) enables machines to learn from previous
human experiences through data inputs and enables continuous learn-
ing from new sets of input data. The development of mathematical algo-
rithms has led to the marked creation of ML, and subsequently to the AI
revolution today.
In this Artificial Intelligence for Risk Management book, we use AI to
mitigate risks through various case studies that will help the reader under-
stand and benefit. AI produces effective and dramatic results in business.
vi Abstract

Many organizations desire to understand and improve risk management


skills to improve their chances of handling risk.
Recently, risk has become important everywhere because of the large
volume of data, different velocity, and variety of data. These aspects of
life appear to be growing bigger and more frequent, often accompanied
by negative impacts. People can use an undetermined amount of risk to
strengthen their position. The range and breadth of risk creates havoc
everywhere in the world, and on a variety of projects. Risk management
is important in an organization because without it, the organization may
have trouble defining its objectives. However, the most important strat-
egy implementation is fear of financial loss. This book focuses on problem
statements with appropriate use cases and proposes AI solutions using
data science and machine learning approaches.
In this book, we hope to provide concrete answers to the crucial
questions facing so many organizations: Where are these risks and what
can be done to lower their impacts? Is AI part of the answers to the risk
mitigations?

Keywords
project management; construction management; program management;
skills development; risk; artificial intelligence; analytics; ML; mitigation;
performance review; data science, and business intelligence
Contents
Acknowledgments�����������������������������������������������������������������������������������ix

Chapter 1 Introduction���������������������������������������������������������������������1
Chapter 2 Introduction to Artificial Intelligence�������������������������������7
Chapter 3 Introduction to Risk�������������������������������������������������������37
Chapter 4 Knowledge Base��������������������������������������������������������������63
Chapter 5 AI Solutions for Risk������������������������������������������������������67
Chapter 6 Identify Risk or Threat Model����������������������������������������85
Chapter 7 Risk Categorization/Classification Model�����������������������97
Chapter 8 Predicting Risk Impact Score Model�����������������������������109
Chapter 9 Risk Probability Occurrence Model������������������������������123
Chapter 10 Risk Priority Model������������������������������������������������������129
Chapter 11 Conclusion�������������������������������������������������������������������131

References�������������������������������������������������������������������������������������������135
About the Authors�������������������������������������������������������������������������������145
Index�������������������������������������������������������������������������������������������������147
Acknowledgments
Our artificial intelligence (AI) and risk expert colleagues across the world
have worked in the corporate world to acquire knowledge and expertise.
AI and risk researchers and practitioners have had a profound impact on
our thinking and on the contents of this book. Subsequently, working
with many corporate customers and serving them has greatly influenced
the preparation of this material.
Thanks to our parents, Godson Addo, Mary (Vanderpuije) Addo,
Seetharaman Centhala, Thulasi Centhala, K. Shanmugam, and S. Saroja,
for encouraging us to get an education and work diligently in the field of
our work. We do not think we would have made it this far without their
support.
We would like to extend great thanks to our families—Louvaine
Addo, Mala Srini, Kavitha Muthu—and Archie’s and Srini children:
Koushik Seethula, Shashank Seethula, Srima Seethula—Muthu’s chil-
dren: Maanasa Muthu, Sarvesh Muthu—for their collective patience with
our busy schedules.
Thanks to Editide for copyediting. We would like to extend great
thanks to Venkat and everyone at Bizstats Technologies Pvt. Ltd. who
enabled us to be Software-as-a-Service (SaaS) BizStats.AI company, which
provided one of the major inputs to write this book.
CHAPTER 1

Introduction

• Target audience
• What you can get from this book?
• What this book covers
• This book’s mind map
• Organization of chapters
ºº Introduction to Artificial Intelligence (AI)
ºº Introduction to Risk
ºº Introduction to the AI Knowledge Base
ºº Business Use Cases
ºº AI Solutions for Risk
• Conclusion
• References

Chapter Outline

• Book Introduction
• Organization of the Book
• Chapter Introduction

Key Learning Points


• Learn and understand: introduction

Target Audience
This book mainly focuses on artificial intelligence (AI) and how manag-
ers can apply AI to risk management. This book follows current trends
in AI in the branch of Natural Language Processing , Natural Language
2 Artificial Intelligence for Risk Management

Question and Answering System of AI, Conversational AI in risk domains,


AI supporting drones, AI cybersecurity, Internet of Things (IoT) devices,
and use cases.
Each applicable AI topic targets:

• Corporate top executives, founders, Chief Technology Offi-


cers, Chief Information Officers, Chief Data Officers, Chief
Security Officers, Chief Risk Officers, data scientists, data
architects, AI designers, AI engineers, project managers, and
consultants to understand how to manage risk using AI.
• Students, teachers, and developers will find this book useful
and practical. It will provide an overview of many AI compo-
nents, and introduce how it can be used in corporate environ-
ments, start-ups, large-, medium-, and small-sized companies.
• Anybody who strives to understand how AI can be used for
risk.

What Can You Get From This Book?

• Understand and learn about AI and how to apply AI to risk.


• Design and apply knowledge-based AI solutions to solve
risk-related problems.
• Architecting and designing AI applied systems that mostly
rely on the following:
ºº Subject Matter Experts. Those with a practical view of how
solutions can be used, not just developed. Here, risk pro-
vides an example using case studies in the book.
ºº Appropriate applied mathematics and algorithms are used
in the book. Do not skip the mathematical equations if you
have the need to study them. It is important to note that AI
relies heavily on mathematics.
ºº Applied physics and usage into hardware systems and
futuristic approaches from quantum computers to parallel
processing of networks in quantum computer handlings. AI
is still evolving with many new areas of possible opportuni-
Introduction 3

ties. Give your full attention to new concepts and applied


creative ideas in the Futuristic AI chapter.
• Decision theory, decision-making process, the Markov Deci-
sion Process algorithm.

What This Book Covers


This book covers mainly an introduction to AI and how it is applied in
corporations, start-ups, large-, medium-, and small-sized companies, to
help automate the tedious jobs of mitigating risk. This book will help
those in organizations’ working environment (as a resource), applying
and automating AI and ML to help human experts.

Organization of
What this book cavers
chapters

Book’s mind map Conclusion

What do you get References

Introduction

AI for risk

Introduction to
Introdution to risk
Artificial Intelligence (AI)

What in intelligence Human senses and AI Definition of risk

What is AI Challenges of AI Risk business case

Types of AI Define AI project Different type of risk

Purpose of AI Risk management


Machine learning process

Risk categories
Unsupervised Supervised

Transfer learning Semi-supervised

Markov decision Deep learning


process
Reinforcement Q-learning
learning
Deep Q network

Figure 1.1  Mind map of the book


4 Artificial Intelligence for Risk Management

Evaluation steps Data collection

Define risk Design algorithm


identification steps
Data preparation
Define goal
Test the model
Identify risk or
Evaluate the model
threat model
Model conclusion

Risk priority model Risk categorization/


classification model
Define goal for Define goal for risk
risk priority model categorization model

Data calculation AI Evaluation steps


for risk priority Solutions Collect risk
Design algorithm for risk category data
for risk priority Design algorithm for
risk categorization
Train the model
Train / test validate
the model
Test the model
Publish/production
of the model
Evaluate the model
Risk probability predicting risk impact
Publish the model occurrence model score model
Define goal for risk Design algorithm Define goal for predicting
probability impact model (risk probability occurrence) risk impact score mode

Train the model Evaluate: DPI,


Test the model
Business process
publish/production
Design alorithm
of the model

Data preparation

Data preprocessing

Test the model

Publish/production
of the model

Figure 1.2  Mind map of artificial intelligence solutions for risk

• How to get true value from AI?


• What are the visionary business use cases for AI?
• How do I identify the best business case to adopt AI and
evaluate opportunities?
• Should I build or buy an AI platform?
• How do I find and recruit top AI talent for my enterprise?
• How will I bring AI into my business to increase revenue or
decrease costs?
• How can I facilitate AI adoption in my organization?

Especially when dealing with data that include data collection, data
preparation, data transformation, securing the data, using the data to
align organizational AI, use cases, and much more. Figure 1.1 provides a
Introduction 5

mind map to give the reader an idea of what is covered in this book and
the organization of the chapters.
Figure 1.2 provides a mind map of AI Solutions for Risk. This figures
gives the reader a sense of which risk areas are covered with AI solutions
in this book.
Index
AI. See Artificial intelligence CNN. See Convolution neural
Artificial intelligence (AI) network
capabilities, 9–10 Contract risk, 48
characteristics of, 8 Convolution neural network (CNN),
data to, 20–21 26–27
definition of, 8–9 Copeland, Jake, 8
distraction of, 17–20 Corporate risks, 48–49
horizontal, 11 Country risk, 48
mind map of, 3–5 Country-specific risks, 46
purpose of, 11
strong, 9 Data
superintelligence, 9 to artificial intelligence, 20–21
target audience, 1–2 definition of, 21
types of, 9 Data collection
vertical, 11 risk matrix and risk trend chart,
weak, 9 81–82
Artificial intelligence (AI) adoption risk register, 81
mass unemployment, 18 samples, 82–83
Data science, definition of, 21
total human control, impossibility
Decision tree algorithm, 35
of, 18–20
Deep learning, 25–27
Artificial intelligence (AI) projects
Deep Q network, 33–34
big data ecosystem, 14–15
Departmental risks, 47
business value vs. time, 13–14
definition of, 12
Eager learners, 35
negative risk of, 15–16
positive risk of, 16–17 Fully connected neural network
value proposition, 12–13 layers, 26
Artificial intelligence (AI) solutions Functional area risks, 47
background, 68
data collection, 80–83 General risks, 57–58
project plan, 68–73
purpose, 68 Horizontal artificial
risk process life cycle, 73–80 intelligence, 11
Human senses, 10–11
Big data ecosystem, 14–15
Business risk, 48 Industry-specific risks, 46–47
Business value, 13–14 Information technology
risk, 48
Classification, definition of, 34–35 Instance-based learning, 23
Classification modeling, 35 Intelligence, 7. See also Artificial
Closed risks, 81 intelligence (AI)
148 Index

Knowledge acquisition, 65 Reinforced learning, 28


Knowledge identification, 65 Risk(s)
Knowledge sharing, 65–66 categorization, 58–61
concerns of, 38–39
Lazy learners, 35 definition of, 38
Liquidity risk, 48 general, 57–58
identification of similarities, 60–61
Machine learning (ML) response, 58
categories of, 22–23 root causes of project, 59–60
definition of, 22 types of, 46–50
reinforcement, 28 Risk business case studies
semisupervised, 25 alternatives and analysis, 44–45
supervised, 23–24 Blockbuster and Netflix, 39–40
teaching machines, 22 Enron case, 42–43
transfer, 28 goals/scope, 44
types of, 22–23 high-level business impact, 44
unsupervised, 24–25 risks/issues, 44
Markov decision process (MDP), Taxi and Uber, 41–42
28–30 Risk categorization/classification
McCarthy, John, 8 model
MDP. See Markov decision process accuracy vs. epoch graph, 103–104
ML. See Machine learning conclusion, 107
Model-based learning, 23 confusion matrix, 102
Monte Carlo (MC) algorithm data preparation, 100–101
crude method, 31–32 defining goal, 97
definition of, 30 design algorithm, 99
simulation, 30–31 evaluating business process, 98
uses of, 32–33 evaluating given dataset, 98
evaluating model, 102–107
Natural risks, 46 evaluating project-related
documents, 98
Open risks, 81 evaluation measure, 97
evaluation steps, 97–98
Project plan model conclusion, 107
background, 68 multiclass classification model,
budget, 69 99–100
constraints, assumptions, risks, and precision–recall curves, 103
dependencies, 69–70 publishing/production of model,
current state of risk process, 70 107
goals and scope, 69 risk category dataset, 99
key stakeholders, 69 testing model, 102
milestones, 69 training model, 101
proposed to goal state, 70–71 tuning model, 101
risk processes, 71–73 weighted majority rule ensemble
Project risk, 48 classifier, 104–107
Risk identification/Threat model
Q-learning, 33 analytical step of evaluation, 87
Q-value value network, 34 binary classification models, 90
Index 149

conclusion, 95 description of, 50–52


data preparation, 91 development of, 52–53
design algorithm, 89 icons of, 53
evaluating business process, 86–87 input and output process, 54
evaluating given dataset, 87 negative risk (threats) response
evaluating model, 93–94 strategies, 56–57
evaluating project-related positive risk (opportunities)
documents, 87 response strategies, 55–56
evaluation measure, 86 project diagram, 53
model conclusion, 94 Risk management standard, 61–62
POS transaction dataset, 89 Risk matrix, 81–82
publishing/production of model, Risk mitigation strategy, 81
94–95 Risk priority model
receiver operating characteristic conclusion, 130
curve, 93 data collection, 129
risk data collection, 88 defining goal, 129
testing model, 92 design algorithm, 129
training model, 91 evaluating model, 130
tuning model, 91–92 publishing/production of model,
Risk impact score model 130
conclusion, 122 testing model, 130
data collection, 111 training model, 129
data preparation, 113–116 Risk probability occurrence model
deep neural network architecture, conclusion, 128
116–117 data preparation, 126–127
defining goal, 109 data preprocessing, 127
design algorithm, 112 defining goal, 123
evaluating business process, design algorithm, 125
109–110 evaluating business process,
evaluating given dataset, 110 123–124
evaluating model, 119–122 evaluating given dataset, 124
evaluating project-related evaluating model, 127–128
documents, 110–111 evaluating project-related
evaluation measure, 109 documents, 124
evaluation steps, 109–111 evaluation measure, 123
model conclusion, 122 evaluation steps, 123–124
Pearson correlation scatter plot, 114 identifying features, 125–126
publishing/production of model, model conclusion, 128
122 publishing/production of model,
regression models, 112–113 128
testing model, 119 regression models, 126
training model, 116–118 testing model, 127
Risk knowledge base objectives, training model, 127
64–65 Risk process life cycle
Risk knowledge development, 64 identifying, 74–75
Risk management process implementing, 78–79
Blockbuster and Netflix-case study, monitoring and control, 79–80
54–55 planning, 73–74
150 Index

qualifying, 75–76 Supervised machine learning


quantifying, 76 definition of, 23
responding, 77–78 linear vs. nonlinear, 23–24
Risk register, 81
Risk trend chart, 81–82 Talent risk, 49
Technology risk, 48
Samuel, Arthur, 22 Transfer learning, 28
Security risks, 49–50
Semisupervised machine learning, 25 Unsupervised machine learning,
Start-up risks, 49 24–25
Strong artificial intelligence, 9
Subject matter risks, 47–48 Vertical artificial intelligence, 11
Superintelligence artificial
intelligence, 9 Weak artificial intelligence, 9

Das könnte Ihnen auch gefallen