See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/346492352

Artificial Intelligence in Engineering and Society - Current Trends and

Applications

Article · November 2020

DOI: 10.12903/DHBW_RV_FN_03_2020

CITATIONS READS

2 1,631

1 author:

Atheer L Salih
Duale Hochschule Baden-Württemberg Ravensburg
22 PUBLICATIONS 670 CITATIONS

SEE PROFILE

All content following this page was uploaded by Atheer L Salih on 23 February 2021.

The user has requested enhancement of the downloaded file.

SCHRIFTENREIHE DER FAKULTÄT FÜR TECHNIK
DER DUALEN HOCHSCHULE BADEN-WÜRTTEMBERG
RAVENSBURG

2020/03

Artificial Intelligence in Engineering and Society –

Current Trends and Applications
Prof. Dr.-Ing. Jens Christian Timmermann, Dr.-Ing. Atheer L. Salih Al-Tameemi
SCHRIFTENREIHE DER FAKULTÄT FÜR TECHNIK
DER DUALEN HOCHSCHULE BADEN-WÜRTTEMBERG
RAVENSBURG
2020/03

Artificial Intelligence in Engineering and Society –

Current Trends and Applications

Prof. Dr.-Ing. Jens Christian Timmermann, Dr.-Ing. Atheer L. Salih Al-Tameemi

IMPRESSUM

Schriftenreihe der Fakultät für Technik

der Dualen Hochschule Baden-Württemberg Ravensburg

Herausgeber

Prof. Dr. Heinz-Leo Dudek

Prorektor und Dekan der Fakultät für Technik

Duale Hochschule Baden-Württemberg Ravensburg

Baden-Wuerttemberg Cooperative State University
Marienplatz 2
88212 Ravensburg
Deutschland

http://www.ravensburg.dhbw.de

2020/03, November 2020

ISBN 978-3-945557-09-9
ISSN 2199-238X
DOI 10.12903/ DHBW_RV_FN_03_2020_TIMMERMANN_AL-TAMEEMI

© Timmermann, Al-Tameemi 2020

Alle Rechte vorbehalten.

Der Inhalt der Publikation wurde mit größter Sorgfalt erstellt. Für die Richtigkeit, Vollständigkeit und
Aktualität des Inhalts übernimmt der Herausgeber keine Haftung.

Druck und Verarbeitung

Gestaltung
Nicole Stuepp
DHBW Ravensburg
Marienplatz 2, 88212 Ravensburg

Druck
Online-Druck GmbH & Co. KG
Brühlstraße 6
86381 Krumbach
Artificial Intelligence in Engineering and Society –
Current Trends and Applications

Prof. Dr.-Ing. Jens Christian Timmermann1,*

Dr.-Ing. Atheer L. Salih Al-Tameemi2

* e-mail: [email protected]

Keywords:
Artificial Intelligence (AI), speech recognition, image processing, robotics, autonomous
systems, smart devices, communication networks.

ABSTRACT

Artificial Intelligence (AI) is a modern field of research that pushes recent technological
developments in different fields of industry. AI fundamentals have origins in mathematics,
information technology and engineering. The scope of this article is to provide an
understanding of basic AI ideas, the evolution of that field and important theoretical
modelling approaches. Furthermore, the paper gives an overview of AI based applications
in science, society and engineering. Hereby, a special focus is laid on some recent
developments in mobile communications and networks. In a final step, the contribution
discusses future trends in AI. In this context, the authors also present an own idea, namely
a hybrid device based on speech and image recognition.

1 Manager of B.Eng. study program “Communications with focus on traffic systems“

within Electrical Engineering, Baden-Wuerttemberg Cooperative State University, Faculty of
Engineering, Germany, 88045, Friedrichshafen, Fallenbrunnen 2
2 Academic staff member Electrical Engineering, Baden-Wuerttemberg Cooperative State University,
Faculty of Engineering, Germany, 88045, Friedrichshafen, Fallenbrunnen 2
1
1 INTRODUCTION

Cambridge dictionary defines Artificial Intelligence (AI) as the ”study of how to produce
machines that have some of the qualities that the human mind has, such as the ability to
understand language, recognize pictures, solve problems, and learn.” (Cambridge
dictionary website, 2020).
AI is a technology leading to advantages in many fields, e.g. by optimizing processes and/or
the processing time. First applications were e.g. found in computer science, where AI
helped to improve software and enabled progress in engineering applications. For example,
systems could react automatically in a clever way when e.g. environmental conditions
changed. AI has been a new scope of research in computer science, especially in the last
two decades.

EVOLUTION OF ARTIFICIAL INTELLIGENCE

Simply, AI is a technology aiming at transferring the human actions and behavior to
machines, imitating rational actions, and thinking. Usually, this process depends usually on
previous input data related to the intended actions. For this purpose, professionals and
experts in particular fields define some goals, and the AI may learn and establish a set of
rules in order to take the right decision. AI has to solve a dedicated problem, and to find an
adequate solution.
There were some first trials to apply AI during the 1940s and 1950s. During the second
world war, there were also first applications of AI in the field of coding/decoding by Alan
Turing and he published his article (Computational Machinery and Intelligence) that
attracted attention and admiration (Muggleton, 2014). In the 1950s, John von Neumann
supported the idea of computer design based on the human brain model (Parkes and
Wellman, 2015; Nilsson, 2014). The first conference on AI was the Dartmouth Conference
which took place in the summer of 1956. This conference has established AI as a new
discipline at that time. McCarthy chose the term ‘artificial intelligence’ to the funding
application of the ‘Rockefeller Foundation’ for this conference. Initially, McCarthy et al.
submitted a proposal of such an AI conference gathering establishment from August 1955
(McCarthy et al. 2006).

2
One of the earliest text books about AI was written by Elaine Rich (Rich, 1983). This book
provides a very good clear primary introduction to the problems and techniques in AI. It
covers important descriptions for some basic techniques in AI problems solving. In 1993,
Ross Quillian was the scientist who instructed a program for a complicated AI net involving
computer simulation of human memory, and language processing. He represented a
semantic network in a graph that consists of nodes and arrows for link connection between
this net (Crevier, 1993).

A well-known AI example was IBM’s Deep Blue chess computer that played and beat chess
master Garry Kasparov in 1997. Later applications involved applications in speech
recognition based on a huge and cumulative amount of data. An important AI revolution
started within the last 2-3 decades in parallel with the revolution of internet and
microprocessors development (Renzi et al., 2014). In the 1990s and early 2000s, AI entered
more and more into the challenge to solve for example problems in information technology
(IT) and business applications (Barzilay et al., 2002; Bartlett et al., 2001; Tennenholtz, 2002).
The field of AI has also become very important in manufacturing and service systems for
the last three decades affecting the daily life (Oke, 2008).

Meanwhile, the AI technology has achieved very high success and popularity in the
scientific community due to the great connection with the development of smart
processors enabling advanced functionalities (Shabbir and Anwer, 2018).
Nowadays, there are a lot of amazing AI results that reduce human effort and impact our
lives in a positive way (such as assisting systems in vehicles with the potential to reduce
traffic accidents and hence the associated death rate). Benefits are not only found in
engineering and industrial fields (e.g. precise robotics), but for example also in medical
applications. In other words, AI can make human life more pleasant and safe (Ashok et al.,
2016). Oke (2008) classifies the areas of AI into 16 areas of application: Programming,
artificial life, data mining, distributed AI, expert systems, machine learning, and neural
networks are considered the most important areas according to Oke (2008).
To summarize, the success of AI can be explained by two steps: First, important basics and
principles of AI were developed showing its big potential and the expected advantages
(Brooks, 1999; Patel-Schneider and Sebastiani, 2003; Zanuttini, 2003). This led to a growing

3
worldwide interest and to an ongoing increase of investments in AI technology. Applying
an AI strategy often optimizes the performance and helps to achieve a deeper
understanding of the considered discipline. (Rosati, 1999; Kaminka et al., 2002; Bod, 2002;
Acid and De Campos, 2003; Walsh and Wellman, 2003). On the other side, there are still
many challenges in the AI field related to unexpected behavior.

GOAL OF THIS PAPER AND APPROACH

The goal of this paper is to present the main ideas of AI as well as current applications and
developments in different fields of science, society, and engineering involving trends. For
this purpose, current references are considered. The paper shows how our everyday life is
already impacted by AI, how mankind already benefits from this development, and which
steps and further applications may come in the future. In this context, the authors also
discuss the idea of an advanced device that could help to preserve a human’s personality.

The paper also provides a good basement and starting point for people and engineers
interested in technological trends. The paper itself discusses several relevant fields,
involving both applications in engineering (e.g. robotics, communication systems), and in
science and society. By understanding the state-of-the-art and upcoming trends in the year
2020, this may help to make technical decisions, but also to understand the impact of AI in
private life.

4
2 THEORETICAL BACKGROUND AND MODELLING
APPROACHES

Nowadays, AI has sometimes become a catchphrase for technologies that realize a clever
behavior. For example, AI can take over more control in complex systems leading to less
end-user interactions and/or increased efficiencies in processes and implementations. To
achieve an AI behavior, the following approaches are typically applied: probability-based
methods, logic principles, numerical methods, and mathematical optimization (Ashok et
al., 2016; Shabbir and Anwer, 2018). This chapter provides a short overview of some
theoretical background and modelling approaches related to AI, without claiming to be
complete as the field of AI is wide.

In order to judge if AI should be applied in an existing classical model (i.e. reference model)
for better performance, several steps have to be considered: In general, the first step is to
explore and study the area of the business domain and the expansion opportunities. This
means to understand the requirements to build up an AI-based system. The next step is to
check the available information, data and the current classical method applied. The
decision of developing an AI system according to the user requirements is the next step
forward. The final step is to face and solve the technical problem based on AI and to handle
the possible unintended behavior. The basic steps of a general AI system are shown in Fig. 1
(Jim Whitehurst, 2019).

Figure 1: AI basic steps

When a reference model is modified by an AI approach, a lot of information and data is

required as a necessary input for the modelling process. After developing the AI model, its
performance is compared to the reference model for further optimization (Shabbir and
Anwer, 2018). During the development phase of the AI model, dedicated rules have to be
defined for decision making.

5
AI is applying three tasks within its formation to generate results: identification, prediction,
and execution. These results obtained in different applications are often much better
performant compared to those of classical models. (Otani et al., 2017). The tasks can be
briefly explained in Fig. 2:

Figure 2: The major stages of AI system formation

To solve a dedicated problem by application of AI, suited AI approaches have to be chosen.

The task is to apply an AI method leading to an optimized performance which may be even
better than achievable by human interaction or human thinking. In a modern project, even
multiple AI approaches may be applied simultaneously (hybrid system) such as machine
learning, neural networking, expert systems, etc.

The following sections present some important terms related to AI that are often used in
literature.

2.1 EXPERT SYSTEM

An important term related to AI is the Expert System (ES), which simply means that a
computer software system based on human expertise to make decisions and deal
with/solve problems has either a high degree of knowledge complexity or incomplete
information. (Uraikul et al., 2007; Oke, 2008). For example, control technology normally
applies an ES as it very useful in active processes. The applied so-called rule-based
6
knowledge which requires deep understanding of the inputs decides the action based on
the actual situation. Expert systems typically use the following items: knowledgebase, user
interface, database, reasoning machine, machine interpretation, and knowledge
acquisition (Uraikul et al., 2007).

An early study related to AI and ES was done by De Toni et al. (1996) that concerned the
production scheduler. De Toni et al. applied a hybrid multi-stage (production and
inventory) ES technology approach to schedule the push/pull orders and the results were
encouraging (De Toni et al., 1996). Exemplary references related to AI expert systems in
imaging applications can be found in (Varela et al., 2016.; Chen et al., 2016) and in (Shabbir
and Anwer, 2018).

2.2 MACHINE LEARNING

One of the main AI fields is Machine Learning (ML) as these fields are often cooperating
together in a lot of applications such as data processing, image recognition, security, and
search engines (Portugal et al., 2018). ML is for example often used to recognize patterns
in data. This learning is based on some reference knowledge and leads to dedicated skills
such as detection or classification. These skills improve over time resulting in reliable results
(Goldberg and Holland, 1988) when manual interaction with a programmer is applied.

There are a lot of studies and research about ML that serve AI technology, especially in the
last three decades (Gamberger and Lavrac, 2002; Meek, 2001; Walker, 2000). One of the
earliest ML studies was done by Soderland and Lehnert (1994) where a system was built
for acquiring knowledge in high-level information extraction processing and the results
were well comparable to the manual ones (Soderland and Lehnert, 1994). Another study
by Blockeel et al. (2002) focused on the development of a new inductive logic programming
based on the query packs application to improve the efficiency.

The ML field is widely connected to other fields, for example, ML takes the leading role in
solving the so-called knowledge acquisition (KA) bottleneck through applying expert
systems (Goldberg and Holland, 1988). Originally, the knowledge acquisition (KA)

7
bottleneck problem started from the fact that the experts are really good at data collecting
and archiving but they have some weaknesses in presenting their experience in a qualified
production rule (Wong and Wang, 2003). ML methods are applied to extract the knowledge
from data automatically and to solve the bottleneck problem (Bakshi and Stephanopoulos,
1994).

Within the start of the current century, a revolution of ML-based studies and researches
shined more in the industrial field (Oke, 2008). In 2002, Nock (2002) presented a concluding
voting classifier based on his theoretical results, approximation algorithms and
experiments. The obtained results were based on a weak learning framework which
resulted in simplicity and accuracy (Nock, 2002). Lerman et al. (2003) had another
important study regarding ML which carried out the Wrapper maintenance problem
through presenting an efficient algorithm that extracts data from positive structural
information examples only (Lerman et al., 2003). A wrapper function is a computer
software application used to extract information from a semi-structured source and to
convert it to a structured format with little or no additional computation.

A further application of ML is the field of adaptive agents: In his study, Gordon (2000)
presented developed ML-based agents to obtain a higher degree of efficient re-verification
after learning with a fast response time. The developed system was very adaptive,
prospective, and precise (Gordon, 2000). So-called reinforcement learning algorithms were
an interesting study done by Brafman and Tennenholtz (2003): for example, in gaming
applications where dedicated goals can only be achieved when multiple players cooperate
in an intelligent way, the mentioned authors could achieve very good results and solutions
through applying simple model-based algorithms (Brafman and Tennenholtz, 2003).
In general, ML became very interesting and popular due to the increased research in the AI
field as a lot of researchers had the idea of how machines can learn from data. Since the
1990s, ML has become more and more popular worldwide (Langley, 2011). However, some
researchers insist that ML is a sub-field of AI but some others deal with AI and ML as
separate fields.

8
2.2.1 DEEP LEARNING
Originally, the so-called deep learning concept was derived from Artificial Neural Network
(ANN) research and now forms a part of ML (Lecun et al., 2015). Details related to the term
ANN are explained in a section 2.4. The term deep learning can be defined as the deep
structured learning typically involving neuronal networks so that the learning process
improves without interaction by a programmer. It is a very modern part of AI and based on
understanding and learning the intrinsic structure of data (Zhong et al., 2016). This leads to
so-called data representation. (Alpaydin, 2020). Deep learning exists in three general
forms: supervised, semi-supervised, or unsupervised which is not considered in more detail
in the frame of the paper. Instead, some applications related to deep learning are shortly
presented:

When studying the processes of a biological nervous system, it can be found that these
processes can be modelled by deep learning approaches that help to explain the
relationship between different stimuli and the related neural responses in the brain. This
involves the interpretation of information processing and communication patterns through
applying a neural network. (Uraikul et al., 2007).

Recently, deep learning has also been applied in so-called big data applications. Hereby,
the idea is to form new efficient systems that learn data representations from unlabeled
data sets and large-scale data. Deep neural networks, deep belief networks, and recurrent
neural networks are considered as the newly developed versions of deep learning. These
new forms allow for a wider range of fields to be applied such as for example computer
vision, natural language processing, speech and audio recognition, and mobile
communications. Deep learning is usually providing a result which is comparable to what
can be achieved by human experts and professionals.

2.3 PATTERN RECOGNITION

Pattern recognition is defined as an automated process of recognizing patterns in data
which means for example to find some kind of repetitions, similarities and regularities. As
an example, pattern recognition is applied to identify cars in sensed video data, and the
9
technology is also often used in monitoring applications. Pattern recognition applies a data
classification process. Criteria have to be defined for this classification while it can be
distinguished between a knowledge-based approach (involving for example object
dimensions to be found for identification purposes) and a statistical approach (statistical
criteria to be defined). Some methods of pattern recognition applying ML or an ANN are
presented in (Uraikul et al., 2007).

There are a lot of applications based on pattern recognition. The following paragraph
briefly presents an example aiming in understanding human intuition by means of AI
pattern recognition. In Frantz (2003), the author describes his ongoing research w.r.t. the
question if human behavior such as decision making, way of thinking and creativity is just
mysterious or analytic and describable by pattern recognition. Frantz (2003) performs a
comparison study between his own findings and those of the expert Herbert Simon.
Herbert Simon presents a lot of studies covering different fields such as AI, decision theory,
psychology, cognitive science, and organization theory, and these studies are based on the
human mind and human thinking (Frantz, 2003).

To summarize: The pattern recognition field is basically focusing on and explaining the
requested pattern, and ML tends to focus more on increasing the recognition rates. Both
of them cooperate a lot and are increasingly involved in AI and statistical approaches. The
pattern recognition algorithm can be divided into either supervised or unsupervised or in
another way to either statistical or non-statistical.

2.4 NEURAL NETWORK

The neural network simply means a network of electrically built cells called neurons. A
modern subgroup of a neural network is an Artificial Neural Network (ANN) which consists
of artificial neurons (Hopfield, 1982). The original idea of neural networks was to imitate
the processes occurring in the human brain by using some microchips. These microchips
consist of artificial neurons that are internally connected and that communicate with each
other for data exchange. In general, a neural network can be distinguished as either a
‘biological neural network’ which deals with biological neurons or an ANN which is used as

10
an AI solution. The ANN is a computational structure of artificial neurons that are
interconnected and arranged in layers and provide the connection between the inputs and
the outputs of the system through certain algorithms (Ashok et al., 2016). The
interconnections between artificial neurons and the type of network topology are the key
elements influencing the ANN behavior (Hassoun, 1996). Modern developments in the field
of ANNs are for example so-called Generative Adversarial Networks (GAN).

There are a lot of ANN-based approaches to apply AI technology, especially in fault

detection, diagnosis, and control systems. An important study was written by Cubillos and
Lima (1998) where they presented a hybrid system using knowledge and a neural network
to implement control strategies based on the related so-called fuzzy parameters. Another
study was presented by Tsai et al. (2003) that investigated process control in chemistry
applications. The authors achieved to model a predictive control architecture and analyzed
dynamic artificial neural network models. Another interesting study was done by Power
and Bahri (2005) as they presented a two-step supervisory fault detection system based on
neural networks, to identify the exact location of faults to solve the problem with high
efficiency.

A neural network can be found in different applications. Four exemplary groups of

applications are considered hereafter: The first group is Character Recognition where the
neural network is used to recognize handwriting as a useful example. The second type of
neural network-based applications is Image Compression and Recognition as the neural
network utilizes and processes the piles of information data leading to encouraging results.
Stock Market Prediction is the third group of neural network applications as the neural
network method can use the market information and history to predict the right time of
fall/rise price and the amount of product consumption. Finally, there are a lot of successful
solutions given by neural networks in the fourth group of applications, namely Medicine
and Security systems (Ashok et al., 2016).

11
2.5 GENETIC PROGRAMMING (GENETIC ALGORITHM)
A genetic algorithm can be defined as a search algorithm based on a certain mechanism
that leads to natural selection comparable to processes in genetics. Genetic algorithms are
considered as a major field of AI. The procedure of genetic algorithms is repetitive in order
to keep certain structures as candidate solutions for the problem of interest. Each
generated structure in any domain is built based on certain genetic operators (i.e.
reproduction and mutation) and then analyzed for the final assessment based on criteria
such as effectiveness (Oke, 2008).
There is a lot of research concerning the importance of AI and genetic algorithm and their
grown relation between them. Turney (1994) presents a very good study about Inexpensive
Classification with Expensive Tests (ICET) as an ideal example of a genetic algorithm-based
AI approach. In this study, the author investigates an approach to “balance the costs of
various possible medical tests with the expected benefits of the tests for the patient.”
(Turney, 1994). The ICET applies a genetic algorithm to develop a so-called decision tree
induction algorithm, and the author shows a comparison with another four classical
algorithms in regard to cost-sensitive classification (Turney, 1994).

2.6 KNOWLEDGE REPRESENTATION

Knowledge Representation (KR) is a very important approach in modelling knowledge and
providing an easy access to stored data. The basic idea is to develop models that collect
information about a dedicated topic or area of interest in order to solve complex problems
within that field. For example, a KR can help to diagnose a medical condition based on
certain knowledge data. There are even examples where a small knowledge base leads to
highly qualified and powerful services (Franklin, 2003).
Exemplary research related to KR can be found in (Calvanese et al., 1999; Cadoli et al., 2000;
Kusters and Borgida, 2001).

12
3 EXEMPLARY APPLICATIONS IN SCIENCE, SOCIETY, AND
ENGINEERING

This chapter considers exemplary applications of AI in different fields of science, society,

and engineering. This involves, e.g., speech recognition, also in combination with intelligent
home appliances affecting our daily life. Furthermore, image processing is an important
field of application: For example, AI is used in modern cameras, television, object detection,
and image reconstruction. In the medical sector, AI helps e.g. to provide optimal medical
treatment and to model biological processes with better understanding. In the industrial
sector, intelligent so-called predictive maintenance is possible, and logistics can be
optimized. AI is also beneficial in the banking and business field, e.g. supported by
intelligent cybersecurity approaches. AI has also become established in the creative sector
impacting arts, gaming, and music. Last but not least: Many applications of AI are of course
related to engineering in general, e.g. to robotics, energy, and the automotive sector.

3.1 IMPORTANT TECHNOLOGIES

While chapter 2 has considered theoretical approaches related to AI, section 3.1 deals with
important resulting technologies, namely with speech recognition, smart sensing, smart
image processing, classification and predictive maintenance. Of course, these technologies
can be used in different context: for example, image proessing may be used in engineering
applications, but also in medicine. Therefore, the subsequent section 3.2 will then have a
look into dedicated fields such as science, society and engineering.

Speech recognition
Speech Recognition is applied in many devices (such as Amazon’s Alexa, The Google
Assistant or Apple’s Siri). The technologies typically work very well for special applications
based on human instructions (“switch on the radio”, “navigate to ..”, “tell me the weather”,
“put s.th. on the shopping list”). Of course, the idea of a system that understands a deeper
meaning or the sense outside a dedicated task is a challenge (Sivadasan, 2008), but there
is continuous progress and evolution in this approach. Modern applications of speech
recognition are speech assistants in cars (Ingenieur.de website, 2020).

13
Smart sensing in intelligent home appliances
The idea is that advanced home appliances automatically optimize the desired result by the
means of intelligent sensing and/or AI. In many cases, optimizing the required resources is
a necessary constraint. An example is an intelligent washing machine that recognizes the
laundry (e.g. weight, colors) and automatically chooses the correct washing program to
save time and resources (Ingenieur.de website, 2020). Another example is an autonomous
vacuum cleaner which remembers the local distribution of dirt to adapt the intensity of
cleaning in the future (Ingenieur.de website, 2020).

Image processing
An important area of AI applications is Image processing. Some selected applications
involving AI are described hereafter:
An important application is computer vision: Hereby, digital images (or arrays of images)
taken by a camera are analyzed and processed to extract relevant information. Computer
vision plays a significant role in robotics, but also in the field of self-driving cars, enabling
navigation or recognition of objects, persons, etc.
Another example related to AI in image processing is the upscaling of images: The image
resolution is improved by intelligent extrapolation of the available data. Hereby, the actual
image is compared to millions of reference patterns in order to apply a dedicated formula
for upscaling. For example, this procedure is applied to generate a computational 8K
resolution for an image with lower quality (Ingenieur.de website, 2020). An exemplary
contribution related to image upscaling can be found in Sharmila and Leo (2016). Upscaling
is e.g. used in television: In Samsung Newsroom website (2020), Samsung reports about
the capabilities of the Q950’s next-generation Quantum Processor 8K in modern TV,
featuring for example:

• improved AI upscaling based on ML

• Object Tracking Sound+ function, an AI-based technology which “matches the
movement of audio sound with the movement of objects on the screen” (Samsung
Newsroom website, 2020)
• AI ScaleNet function, a method which optimizes the available network bandwidth
to ensure a smooth streaming connection

14
Further applications of AI in the field of image detection and processing are AI-powered
cameras (Analytics India Magazine, 2019): These cameras may for example support
upscaling, a special night mode in a low-light environment, or scene detection which means
that only the best photo is taken out of a set of photos. According to Analytics India
Magazine (2019), exemplary mobile phones with AI cameras are Google Pixel 3, Huawei
Honor 10, Samsung A50, iPhone XS Max, OnePlus 7 Pro, Vivo V15 Pro, and Oppo R15 Pro.
In modern mobile phones such as the Samsung Galaxy S20 (launched February 2020 in
Germany), AI is also used to perform motion analysis for video stabilization (Zou et al.,
2019).
A further application of AI in the field of image processing is face reconstruction based on
a poor-quality image, e.g. taken by a security camera at airports or railway stations (Railway
Technology website, 2017). In the latter use case, this can help authorities to find dedicated
or missing persons.
Another application of AI related to image processing is the possibility to perform
colorization of black and white photos based on deep learning: To train the AI algorithm, a
large set of reference black and white photos is used where the corresponding color
information is available. As a result, the reconstructed colors can be compared to the
original ones. The delta between original colors and reconstructed colors is then used to
optimize the estimation process. The website https://demos.algorithmia.com/colorize-
photos offers the possibility to upload private black and white photos and to download the
colorized versions. The applied deep learning algorithm is described in detail in Zhang et al.
(2016). Concurrent work can e.g. be found in Larsson et al. (2016). Fig. 3 shows an
exemplary colorization of a photo: Original and reconstructed colors are very similar.
Colorization of black and white photos can also be combined with an AI-based
reconstruction of a poor quality image: Hereby, details of the original scene are estimated
(e.g. details of a face) although the relevant information is only partly available: Since 2020,
such a service is offered at the MyHeritage website for paying customers, leading to
amazing results.
Finally, AI is also used in image search: By means of intelligent classification, the AI
application may browse the world wide web or a database to provide similar photos based
on a dedicated uploaded photo. For example, search engines such as Google support this
function. Furthermore, in November 2018, the German eBay app has added a function that

15
allows users to upload a photo and eBay will search for comparable products. This function
applies to image recognition based on ML (Ebay Press Room, 2018).

Figure 3: Exemplary colorization of a photo by applying the algorithm as per (Zhang et al., 2016):
three pictures: the original in colors (on top), then a converted version in black and white
(middle photo) as an input to the colorization process, and finally the photo with
reconstructed colors (last photo).

Smart classification of documentation

The idea of smart classification of documentation is that an AI approach reads (e.g.,
unsorted) documents and performs identification and/or categorization. For example,
available clinical data from the past can be sorted (IBM website, 2020), evaluated, and used
to improve medical treatments for similar actual patients. Another example is the legal
environment: relevant judgments for an actual case are found by an AI-based smart
classification system which helps lawyers and judges to get a better idea of strategy.
However, the results have to be checked in detail as they rely on the past and may not be
compatible anymore with recent developments in the case of law. There is also the danger
16
that hidden discrimination of social groups that may have occurred in the past is applied
again by the algorithm (Deutsche Welle, 2019).
Classification of documentation can also be combined with speech recognition: In a first
step, voice commands are recognized and translated into text. In a second step, an AI
algorithm performs a classification of documentation based on the text. An exemplary
application is the production environment: An employee working in the production line
may for example search a dedicated installation guide. Instead of searching it manually, he
speaks to a “smart glass” which finds the required document automatically and may display
its content. (Elektrotechnik Automatisierung website, 2019). A manual search is hence not
required anymore which helps to increase efficiency.

Predictive maintenance
The idea of predictive maintenance is to monitor the health (in the sense of the status) of
a technical system (called condition monitoring) by dedicated parameters, to analyze the
collected data (data analytics), to derive the health status by an AI algorithm, and to issue
a maintenance of the system in case of an anomaly before the system fails. Hereby, a
technical system can be for example a dedicated machine, a driving train, or a production
line. Exemplary parameters for assessment can be the voltage, temperature, pressure, or
acoustic information. For example, anomalies may result in the generation of heat or
additional noise. In the production environment, predictive maintenance helps to both
increase the reliability and to minimize the corresponding downtime/repair time.
(Elektrotechnik Automatisierung website, 2019).

3.2 APPLICATIONS IN SCIENCE, ECONOMY, AND SOCIETY

The subsequent consideration presents some actual applications of AI in science, society,
and economy. Considering science, exemplary applications in the medical sector and
genomics are shown. Regarding the economy, the impact of AI in logistics and banking is
considered. Finally, some applications of AI are described that affect cultural life and leisure
behavior: AI in arts, music, and gaming.

17
Medical applications and healthcare
An exemplary application of AI in this field is the prediction of the spread of a virus: Hereby,
an algorithm automatically collects dedicated information about the actual occurrence of
a virus in reports, networks, etc. to get the actual status. This information is used to feed
prediction models and to generate warning notices for authorities. As a consequence,
countermeasures can be undertaken to limit the spread of the virus (Tech News website,
2020). A well-known algorithm in this field is the so-called BlueDot algorithm. Currently,
the prediction of the spread of the Coronavirus is of great interest. Coronavirus (Covid-19)
started to spread worldwide possibly at the end of the year 2019 and extended more and
more as a pandemic in 2020. Researchers and some scientific institutes started to apply AI-
based techniques to predict where the virus will spread and how long it will take; the idea
is to instruct an early alarm for this threatening virus spread. Relevant AI studies are based
on the obtained information from online sources, medical reports, and the World Health
Organization (WHO) numbers. These studies rely on the SARS virus spread in the 2000s
which killed 774 people all over the world (Time website, 2020).
Another important application in the medical field is the Blue Brain project which has been
founded in 2005 in Switzerland. A result of this project was to model about 11% of the total
human brain (85.9 billion neurons) in a supercomputer based on the AI state of the art
(Markram, 2006; Eliot, 2009). This project has opened the way for further medical studies
and applications based on AI technologies.
Further applications in the field of medicine are for example the detection of skin cancer
by AI-based dermatologic image analysis (Reiter et al., 2009). In the biopharmaceutical
field, AI is applied in drug discovery (Nature website, 2018). Finally, AI is also applied to
derive clinical diagnoses or to suggest treatments. The latter two topics and further medical
applications are described in more detail in (Chan et al., 2018).

Genetics
In the field of genetics, AI has become an important method to understand relevant
processes that help for example in health surveillance and therapy, but also in genealogy.
Science Daily website (2019) reports about research in genomes (genetic material of an
organism) and the discovery of a new hominid that had once lived in the past; this
conclusion was derived by combining both a deep learning and a statistical approach.

18
Emerj website (2019) gives a good overview of current AI applications based on ML in the
field of genetics. In so-called genome sequencing, two strategies can be distinguished: the
first has an emphasis on research to achieve for example progress in individual therapies
and to understand an individual risk of disease. A second strategy is to focus more on the
business case: For example, analysis of individual DNA for paying customers lets them know
about the genetic impact on weight (Emerj website, 2019). Further applications are the
prediction of so-called pattern changes in DNA structures by deep learning (Zou et al.,
2019).
In the field of genealogy, a DNA analysis can help to understand the personal origin, e.g. in
the sense that it is estimated how typical the DNA would be for certain geographical regions
in the world. Hereby, the group of people in a certain region is called an ethnicity. Hence it
can be estimated in which parts of the world the ancestors may have lived. To derive such
a conclusion, individual results are compared to the results of reference persons with
typical patterns of a certain geographical region. By applying this process, the personal
distribution of ethnicities is revealed. By mapping individual results with those of other
people who have done a DNA test, so-called DNA matches can be found. In the case of a
DNA match, the probability is high that there is a common ancestor.
Companies such as MyHeritage or Ancestry offer a dedicated service showing, among
others, both the individual ethnicities and DNA matches. Ancestry Support website (2020)
provides a link to dedicated white papers explaining the applied methodologies. In a typical
case, several thousand DNA matches are found nowadays. Of course, the number of DNA
matches depends on the number of users that have done a DNA test. Furthermore, the
degree of relationship is estimated (e.g., father, mother, uncle/aunt 3rd grade, etc.). As
users can also upload a public family tree, this information simplifies the identification of
the common ancestor. It is also possible to include relevant family trees of other users into
the own tree and to upload and share additional information (e.g. old photos and
documents). All these functions contribute to deepen the knowledge of the family roots.
Typically, a smart algorithm supports the users in finding relevant family branches to be
included into a family tree. This is done by comparing the individual tree to other trees.
MyHeritage service also offers the possibility to improve old black and white photos: AI
algorithms improve the quality, detect faces, estimate missing information, and apply
colorization.

19
Logistics
AI is also applied in logistics. It helps to ensure smooth processes, to optimize cost,
efficiency, safety, and customer satisfaction. Towards Data Science website (2019) gives a
good overview of current applications, projects, and involved companies in the field of AI-
supported logistics. For example, Towards Data Science website (2019) considers
automated warehousing: Hereby, an AI approach estimates the required number of goods
in different warehouse stocks which optimizes the transportation effort between the
warehouses. In a warehouse itself where workers (so-called pickers) have to find ordered
goods, AI can be applied to guide the pickers, to optimize the route and time. SSI-SCHAEFER
website (2018) describes a dedicated approach applied at Zalando company.
Another field of application is autonomous vehicles for good transportation in indoor or
outdoor environments. Inside a warehouse, these autonomous vehicles can be equipped
with appropriate sensors and devices for navigation purposes. If all vehicles communicate
with each other and share relevant information such as actual position and operational
capability, an AI approach can be applied to make decisions based on optimization goals.
For example, the swarm can decide which vehicle will move to a destination of interest
while optimizing the required time or the length of the route (SSI-SCHAEFER website, 2018).
Considering outdoor scenarios, another idea related to logistics is a smart transport route
that supports the driver or the vehicle itself as a function of environmental conditions.
Exemplary ideas described in (Towards Data Science website, 2019) are the provision of
locally generated power or assisting lights as a function of brightness.

Banking
An overview of actual AI methods applied in banking is e.g. given in Der Bank blog website
(2020). According to Lionbridge website (2020), AI in banking is applied in the following
fields:

• Customer analysis and interaction: Hereby, banking customer behavior is analyzed

to offer e.g. a tailored investment portfolio or credit. This can be realized by a
dedicated chatbot.
• Automated trading: This helps to optimize the profit.

20
 • Optimization of banking processes: This application can perform an automatic
compilation of required or missing documents, e.g. for completing administrative
processes. This helps to minimize the workload on the employee’s side.
• Prediction and estimation: This topic concerns the prediction of the default risk on
loans and the estimation of the profit opportunities.
• Fraud detection: Fraud detection means to identify attempts that intentionally
manipulate banking processes. This is a very important application of AI in banking,
e.g. based on ML (Der Bank blog website, 2020).

Arts
AI can be used in arts to generate paintings in a specific style, e.g. by imitating famous
painters based on deep learning. To generate the painting, also robots can be used. An
actual painting robot project is called “Artonomo.us” (Cloud Painter website, 2020). This
robot first selects an image to paint, then applies a dedicated style based on various AI
methods, and finally starts the painting process.

Music
A further application of AI is the composition of songs. For example, Sony’s “Flow
Machines” applies ML and has created the song “Daddy’s Car” based on Beatles songs (Stay
Relevant website, 2017). Another example is the technology developed by the AIVA
company. The AIVA software can compose soundtracks of different styles (Stay Relevant
website, 2017) such as Electronic, Pop, Rock, Jazz, and classical music. Concerning classical
music, tens of thousands of musical scores written by the most popular composers (such
as Mozart, Bach, and Beethoven) have been used to train the algorithm. Applications of
AIVA outputs are for example soundtracks for films and video games. To get a better idea
of the results, exemplary songs have been uploaded on "YouTube". AIVA is also the first
virtual artist whose songs have been registered with an author's rights society (SACEM).

Gaming
According to AITHORITY website (2020), the application of AI in gaming aims at providing a
realistic gaming experience in a virtual environment. This can e.g. be achieved by virtual
characters that behave in an intelligent way, that learn from experience, adapt their
strategies to achieve a certain goal, and/or compete with the player’s skills (Wikipedia
website, 2020). The application of AI in gaming can lead to matching actions, reactions, or
21
responses, e.g. expressed by an emulated human voice. As adaptive behavior is a key
element in nature, games emulating such a world are seen as realistic. The degree of
challenge rises to a higher level when applying such an AI approach. This also leads to a
higher level of satisfaction for the player (AITHORITY website, 2020).
However, AI in gaming is not limited to the idea of simulating a smart behavior. The scenery
itself should appear in a realistic way as well. This can be achieved by an AI deep learning
algorithm that learns from real-world photos or videos (The New Stack, 2020). For example,
real-world driving sequences can be used as training data to generate a virtual ride in a
computer game (The New Stack, 2020). According to Synced website (2020), future games
will collect visual information of the player (such as emotions) as an input to derive a
matched response. Furthermore, classical controllers such as joysticks enabling the
interaction with the virtual scene will be more and more replaced by interaction through
human voice and physical movements.

3.3 ELECTRICAL ENGINEERING, ROBOTICS AND IT

In the field of engineering, there are plenty of applications related to AI. The following
paragraphs focuses on exemplary fields related to electrical engineering and information
technology (IT), namely automotive, energy and power, robotics, and cybersecurity. Basic
ideas and relevant applications are shortly presented.

Automotive
AI is widely applied in the automotive sector. For example, this concerns both the
development of a vehicle itself (e.g., AI is applied in the design and manufacturing process)
and the realized driving behavior. In addition, AI-based simulations can mimic realistic
driving situations and help to test and optimize the driving behavior prior to any physical
test that involves a vehicle on the road.
Nowadays, the development and the production of a vehicle is often supported by AI.
Benefits are, e.g., rapid prototyping (Global Market Insight Website, 2020) and an improved
efficiency in the production line. For example, involved equipment in the production line is
constantly monitored (VHR Company website, 2020), and AI algorithms realize predictive
maintenance in the production line to optimize efficiency, time and cost. Built In website
(2020) mentions an approach based on ML where the normal behavior of a machine is
22
learned as a reference. Built In website (2020) also reports about another application of AI
in the production line: Collaborating robots (so-called co-bots) are able to replace humans
in case they are absent to due illness etc.
As already mentioned, the smart behavior of a vehicle is often realized by AI. For example,
modern vehicles are equipped with many sensors that collect relevant information as an
input to driver-assistance systems. Exemplary driver-assistant systems using AI are driver
behavior monitoring (e.g., eye tracking), road condition awareness, and lane tracking
(Global Market Insight Website, 2020). They help to make driving more safe and enjoyable.
These systems often use AI to make decisions based on the sensed data applied in the
algorithm. As the number of involved sensors is constantly growing in modern cars, it is
also important to efficiently process the resulting “big data” within short time. This is
achieved by AI approaches (Benčić et al, 2018). Benčić et al. (2018) also provide an overview
of AI trends in the automotive sector based on an actual patent analysis.
In case there is no driver anymore (self-driving car), AI plays a more dominant role
compared to state-of-the-art vehicles. The applied AI has not only to assist, e.g. to prevent
human driving errors, but it has to understand the driving situation in a way that road
casualties are rather zero. Hence, the AI does not only mimic human behavior, but the
obtained performance has to be better. Typically, the involved solutions make use of deep
learning to train neural networks (VHR Company website, 2020; Global Market Insight
Website, 2020). In self-driving cars, the observation of the vicinity of a car is a key task and
realized by dedicated technology. According to Teslamag website (2020), e.g. Tesla does
not make use of LiDAR (Light Detection and Ranging) or high-resolution maps to realize self-
driving capabilities but relies on cameras. By continuous learning, false alarms are
minimized over time. The actual Tesla Autopilot engine includes 48 neural networks
(Heartbeat website, 2020), where each network solves dedicated tasks. As Tesla relies on
cameras, computer vision plays a significant role. In 2019, Tesla has e.g. acquired DeepScale
which is a start-up from Silicon Valley involved in Deep Neural Network technologies and
computer vision. Electrek News website (2020) reports that the acquisition of DeepScale
pays off, advances the development of the Autopilot, and has led to a dedicated patent
called “Systems and Methods for Training Machine Models with Augmented Data” in April
2020 (Cooper et al., 2020).

23
Finally, some applications in the field of simulation are considered. As already mentioned,
vehicle simulation is an important element to virtually test the capabilities of AI-based
vehicles. For example, a virtual time-varying scenario can be created involving possible
situations such as passing pedestrians, traffic, etc. Both the generation of the scene and
the situations may be supported by AI as well, e.g. based on deep learning methods. In this
context, the Automotive Artificial Intelligence website (2020) mentions an AI-trained traffic
agent populating the streets. By involving sensor simulation, the software applied in the
vehicle can then be tested in the virtual world (Automotive Artificial Intelligence website,
2020). A good example of such a type of software is recognition software that detects traffic
signs, obstacles, traffic participants, lane changes, etc. (Global Market Insight Website,
2020). An exemplary company involved in traffic and scenario simulation, as well as sensor
simulation, is Automotive Artificial Intelligence (AAI) (Motor Zeitung website, 2019): The
virtual ride also encompasses traffic participants with different types of behavior involving
aggressive and dangerous overtaking so that the software under test can learn from these
situations. It is also possible to simulate different road conditions. By applying simulation-
based virtual rides, much effort can be saved regarding the subsequent physical testing on
the road. This helps to speed up development processes (Motor Zeitung website, 2019).

Energy and power

In the field of energy and power, AI is applied in planning, control, operations, automation,
and analysis.

• As a starting point, planning processes are considered that require information

about energy consumption and energy generation as a necessary input. In the
recent past decades, AI-based models have been widely applied in the forecast of
energy consumption. Basically, AI-based models are separated into either ANN-
based models or co-called Support Vector Machine (SVM) based models. An
innovative grid optimization approach based on an ANN is for example described in
(Wenzel, 2019). The results in Wei et al. (2019) show that the AI-based models are
very robust and comprehensive in the investigated forecast scenarios (called
forecasting horizons). Hereby, a forecasting horizon may be a short-term or a long-
term prediction. According to (Wei et al., 2019), the robustness is achieved due to

24
 an evaluation process between the models' performance in various forecasting
horizons.
Of course, forecasting the energy consumption is not the only application of AI in
the field of power. Slideshare website (2017) presents an overview of further
applications: AI is applied in a variety of fields involving e.g. planning of power
systems, operations, and control.
According to Sivadasan (2008), planning issues involve e.g. an optimized positioning
of a wind turbine, optimization of feeder routing, and a smart reactive power
control. The latter task can be achieved by dedicated placement of compensation
devices to realize the required voltage profile (Bansal et al., 2003).
• In the field of control, items to be controlled are e.g. voltage, voltage stability,
power flow, and load frequency.
• In power operations, AI is e.g. applied to minimize power losses, to perform load
management, and to control Flexible Alternating Current Transmission Systems
(FACTS) (Sivadasan, 2008).
• Slideshare website (2017) also mentions the automation of power systems as an
application of AI: Hereby, AI is e.g. beneficial in fault analysis, predictive
maintenance, and network security.
• According to Slideshare website (2017), AI is even applied in the electricity market,
e.g. to derive optimal strategies for bidding and market analysis. Finally, AI can also
be beneficial in the field of technical analysis, helping e.g. to minimize harmonic
distortions (Sivadasan, 2008).

Robotics
Nowadays, the world sees some millions of operating robots where many of them are
already based on AI approaches. In a production line for modern cars, robots are typically
involved in about 80% of the manufacturing steps with marvelous results (Ashok et al.,
2016).
As already mentioned, computer vision plays a significant role in robotics enabling the
understanding of the environment, safe navigation, and recognization of objects and
persons. RSIP Vision website (2020) gives a good overview of different applications of
robots supported by AI. Industry robots are e.g. applied in warehouses and often perform
so-called “pick and place” tasks (Electrek News website, 2020) or even collaborate with
25
humans (co-bots). In the semiconductor industry, (visual) defect detection of a component
is a further possible task (Electrek News website, 2020). In agriculture, robots assist e.g. in
navigation and proper growing and sorting of agricultural products (Electrek News website,
2020).
There are also AI-based robots designed to look like a human: They support and interact
e.g. with customers in supermarkets (Electrek News website, 2020), with patients in
hospitals (Electrek News website, 2020), or senior living homes (Time website, 2019), and
even with guests in restaurants. In restaurants, chat-bots can be applied to support the
customer in the selection and ordering process and for entertainment issues (World
Economic Forum, 2019). For example, the fast-food chain KFC even applies face recognition
so that information about last orders can be used for the service.
In senior living homes, robots can e.g. be beneficial to bring meals or to alert the nursing
staff in case of an identified danger or problem. For example, a robot with voice detection
capability could be able to understand a call for help (Time website, 2019). Robots in senior
living homes can also be applied for entertainment purposes: If a robot is able to recognize
persons, it could play the person’s favourite music or game. In case the robot can recognize
emotions, this information might be used to recognize the person’s needs. Another
possible function is to utilize a mobile humanoid-looking robot in a videocall: The robot can
replace its displayed face by the caller’s face (Time website, 2019). Such a function is very
beneficial when visits are not possible, restricted, or not recommended, e.g. due to the risk
of an infection (e.g., Coronavirus) threatening a senior’s life.

Cybersecurity
In the field of cybersecurity, AI plays an increasing role in defending systems against
adversaries. On the other side, adversaries might apply AI as well (e.g. for attacking
purposes), leading to a competition at a high level. Brookings Institution website (2020)
and Babuta et al. (2020) provide interesting insights into current applications and dedicated
challenges, involving both technical and political considerations.
According to Brookings Institution website (2020), defending applications typically involve
ML methods to detect threats and to initiate dedicated countermeasures. When this
defending process is done automatically, such an approach avoids time-consuming human
interaction (ZD Net website, 2020) and the decision-making processes. Of course, such an
approach has its pros and its cons: On the positive side, attacks can be stopped faster
26
(Babuta et al., 2020) limiting the possible damage compared to a delayed response.
Furthermore, the application of ML can e.g. also detect variants of malware (ZD Net
website, 2020). On the negative side, when AI takes over more control, systems to be
defended are more vulnerable in case that the AI of the defending system is manipulated
by adversaries. This could be applied in a so-called adversarial ML (Brookings Institution
website, 2020; ZD Net website, 2020). In such a case, altered training sequences may be
used to provoke wrong decisions, e.g. to cause a car accident (Brookings Institution
website, 2020). Another exemplary application of AI in the field of cybersecurity is AI-based
network monitoring: Hereby, users' behavior may e.g. be studied by ML as a reference to
identify relevant changes in the behavior that may cause a threat (e.g. clicking on a phishing
link) (ZD Net website, 2020).

27
4 APPLICATIONS IN COMMUNICATIONS

The relationship between AI and communication networks has a long history although
dedicated applications have become very popular only in recent years. The roots of this
relationship can be found in the 1980s and are linked with ANN theory that became famous
worldwide with its first important applications at the start of the 2010s (Charzinski 2019).
Nowadays, in Germany, there are about 45 million active subscribers within the German
mobile communication network operators such as Telekom, Vodafone etc. (Charzinski
2019). Mobile networks deal with much data traffic and complex connections as there are
about 840 mobile network operators in the world with a lot of challenges to be considered
and solved to realize proper processes. Meanwhile, AI technology has become an excellent
solution to realize modern functions and use cases in these mobile communication
networks.
AI technology has taken a leading role in the revolution of the communication industry
during the last decades as it has helped to add new services and to increase the network
efficiency (Xu et al., 2017). In addition, AI is beneficial to realize autonomy in
communication networks, e.g. to identify anomalies (for example interference) and to
resolve them without human interaction. AI is also helpful to realize the required high data
rates in modern mobile networks, for example by smart spectrum sharing AI approaches.
Also in planning processes of mobile networks, AI is an advantageous method to predict
data traffic (Otani et al., 2017).
The fast revolution of smartphones and the worldwide spread of mobile applications forced
the industry to develop and update the wireless networking infrastructure (Zhang et al.,
2019). The modern 5G systems are specially designed to cover the huge required mobile
traffic volume, to manage the network resources softly, and consequently to be compatible
with the rise of mobile users’ expectations.
The advanced ML techniques are one of the practical solutions to achieve these goals in
mobile environments that are increasingly complicated, diverse, and gradually developed
(Zhang et al., 2019). Zhang et al. (2019) provides a summary of deep learning applications
for mobile devices and systems as a survey paper. In this context, the paper also considers
benefits, limitations, applied techniques within different deep learning approaches, applied
optimization methods as well as architectures related to deep learning. For example, deep

28
learning has the potential to achieve an “appropriate balance between accuracy, latency,
storage and energy consumption for deep neural networks on mobile platforms” (Zhang et
al., 2019). Storage issues typically occur when “big data” streams contain similar content:
A deep learning approach is helpful in identification of these situations.
There are plenty of deep learning-based applications in the mobile networking field, and
the most common advanced applications can be summarized in Fig. 4:

Figure 4: Deep learning – based Mobile Network Systems

Network operation also benefits from AI approaches. Some selected work revealing the
advantages is shortly presented:
Kojić et al. (2006) suggested the optimization of routing in communication networks
through the implementation of neural network technology. In another study, AI technology
was applied to manage routing protocols in communication networks using a Software-
defined Networking (SDN) (Sendra et al., 2017). Another work (Sui and Jin, 2011) presented
an AI-based approach to control the possible reduction of quality of service in a network
(called network congestion). AI technology can also provide a practical solution in
communication networks based on the collaboration between the SDN and the Network
Function Virtualization (NFV) as they behave as AI tools in reasoning, learning, networks
service, and deep packet inspection (Xu et al., 2017). In recent years and parallel with the
29
huge development in AI technology, a lot of communication network operators have
started to implement AI methods due to the ability to decrease capital expenditure and
operating expense (Xu et al., 2017).

As mentioned above, another important aspect in modern mobile communication

networks is the capability to realize high data rates. Increased data rates can typically be
realized by an increased bandwidth, but bandwidth is physically limited and expensive.
Therefore, smart AI approaches apply for example spectrum sharing to cope with minimum
resources. Some theoretical background and an exemplary AI approach are considered in
more detail hereafter:
AI can be used to optimize spectrum sharing in wireless communications between a set of
wireless devices. Traditionally, wireless technologies use fixed, allocated frequency bands
to avoid mutual interference. A better approach is the idea of frequency-hopping where
frequencies are changed over time which can help to exploit the unused spectrum and to
avoid interference. Hereby, an algorithm defines the frequencies over time. However, the
performance of the frequency-hopping approach is limited by the availability of unused
spectrum (Tilghman, 2019). When all involved devices transmit data, the overall achievable
data throughput is not maximized by the frequency-hopping approach. To increase the
overall throughput (optimization goal) while avoiding interference, dedicated approaches
are described in Tilghman (2019): One approach applies dedicated AIs for the different
involved radios, and each AI continuously adapts its own rules to target the overall
optimization goal. Tilghman (2019) describes this process as an “evergrowing series of rules
by mining them from a large volume of data”.

30
5 TRENDS AND FUTURE APPLICATIONS

AI is applied in many fields, and in each field, ongoing progress leads to specific trends and
possible future applications. Some of them have already been considered in the previous
sections. However, for the sake of clarity, the aim of this chapter is not to list all the trends.
Instead, the authors would like to share some ideas and considerations related to an
exemplary field of application, namely the intelligent home appliances.
Having experience with intelligent home appliances currently on the market, some
upcoming ideas were spread among the team of authors that are shared hereafter: As
progress normally never stops, the authors think that it’s rather a question of time when
the next generation’s intelligent home appliances will be available, combining at least
already existing technologies in the field of AI to form a hybrid system. One idea shared in
the team could be called the “preserved personality”:

The vision starts with a device installed at home which is able to recognize a specific person,
e.g. based on speech recognition. This is already possible, e.g. with Amazon Alexa. The goal
of the device could be to learn the specific behavior of a person so that elements of the
personality are preserved over time. By means of AI, imitation of the personality would
become possible, and the device could be applied as a dialogue partner although the real
person is absent. The possibility to hear the voice of a loved one in a known way may be
beneficial for many persons. Furthermore, future generations of mankind could have an
impression of the personality of their ancestors. The device would have to learn specific
elements of a person’s behavior such as speech, typical vocabulary, pronunciation, or
emotional reactions from speech in the context of specific situations.
Such an approach based on speech recognition could be combined with advanced image
recognition (hybrid device): In case the device is equipped with a camera, the device could
perform image detection to recognize different persons and in parallel, it could study the
behavior of each person. The latter case means to learn typical movements and the body
language. For imitation purposes, a virtual person could appear on a display, or a hologram
could be generated during conversations. Dedicated technologies could be applied to
generate the required movements that are synchronized with speech: Relevant algorithms
are already in development. Some relevant background and actual possibilities to
distinguish virtually generated persons and real existing persons are for example described
31
in (Nguyen et al., 2019; Tolosana et al., 2020). To save changes in behavior over time (e.g.
related to childhood, youth, etc.), the hybrid device has to generate several big data sets
that cover the corresponding age groups. Of course, the ability to preserve and imitate a
personality has its advantages and disadvantages:
• For private use, the hybrid device as described above might be beneficial to keep
the memory of family members.
• On the other side, an access to people outside the family is not recommended as
this would offer the possibility to know details about private life based on big data,
e.g. obtained over a long time.

Similar ideas have currently been published for example in (Daily Sabah website, 2020),
describing a “digital twin”, also called a digital avatar.

Last but not least, this chapter summarizes some trends related to AI: According to Forbes
website (2020), more and more AI will be applied in more and more devices in the future.
Some AI will interact with us, but we will not necessarily realize it. Examples where AI
already interacts with people are personalized results originating from search queries in
the world wide web, personal assistants such as Amazon Alexa, or chatbots in digital
customer services. In future, the role of AI will become more dominant in films, music, and
games (Forbes website, 2020): For example, the film “The Irishman” (year of publication:
2019) shows a de-aged Robert de Niro. Hereby, special effects have been applied based on
AI (Forbes website, 2020; Vulture website, 2020).

Exemplary literature describing trends and possible future applications related to AI can be
found in Forbes website (2020) and Techiexpert website (2019).

32
6 SUMMARY AND CONCLUSION

The paper has presented fundamentals related to AI to provide a basic understanding of

classical and modern approaches. Hereby, some modelling approaches have been
discussed. In the next step, the importance of AI technology has been highlighted by
showing a variety of applications in science, society and engineering with some emphasis
on communications. Finally, the authors have suggested a new hybrid device combining
speech and image recognition that can easily be installed at home and can affect the daily
life in a positive way.

As a conclusion, there are versatile applications of AI in both technical and non-technical

context. AI is not only applied in science and engineering, but it also affects e.g. economical
and legal decisions. It even impacts the daily behavior of persons as well as their decisions,
although they may not expect a hidden guidance by AI: Some decisions may already be
taken based on AI-supported search results or the presentation of products on a website.
The behavior and way of thinking may be impacted by content proposed by AI, e.g. in social
networks or streaming portals. And finally, it should not be forgotten that also the date
matching industry already applies AI (Synced website, 2019). These examples show that AI
plays a significant role in modern life. This guidance leads to amazing advantages in
particular fields such as engineering, but the negative side should not be neglected.

ACKNOWLEDGMENT

We would like to express our special thanks to the review team for discussion and feedback.
In addition, we thank our family members for their backing during the last weeks.

33
REFERENCES

Acid, S., & de Campos, L. M. (2003). Searching for Bayesian network structures in the space
of restricted acyclic partially directed graphs. Journal of artificial intelligence research,
18(1), 445-490.
AITHORITY website. (2020). Understanding The Role Of AI In Gaming,
https://aithority.com/computer-games/understanding-the-role-of-ai-in-gaming/ ,
Feb 19th, 2020 ; [accessed July 8th, 2020].
Alpaydin, E. (2020). Introduction to machine learning. MIT press.
Analytics India Magazine. (2019). A nirudh VK. Top 7 Smartphones With AI Cameras And
Post-Processing, https://analyticsindiamag.com/top-7-smartphones-with-ai-
cameras-and-post-processing/ , June 28th 2019 ; [accessed September 7th, 2020].
Ancestry Support website. (2020). AncestryDNA White Papers.
https://support.ancestry.com/s/article/AncestryDNA-White-Papers ; [accessed
September 9th, 2020].
Ashok, J., Suppiah, S., & VijiPriya, J. (2016). A review on significance of sub fields in artificial
intelligence. International Journal of Latest Trends in Engineering and Technology
(IJLTET), 6(3), 542-548.
Automotive Artificial Intelligence website. (2020). https://www.automotive-ai.com/ ;
[accessed July 9th, 2020].
Babuta, A., Oswald, M., & Janjeva, A. (2020). Artificial intelligence and UK national security:
Policy considerations.
Bakshi, B. R., & Stephanopoulos, G. (1994). Representation of process trends—III.
Multiscale extraction of trends from process data. Computers & chemical
engineering, 18(4), 267-302.
Bansal, R. C., Bhatti, T. S., & Kothari, D. P. (2003). Artificial intelligence techniques for
reactive power/voltage control in power systems: A review. International journal of
power & energy systems, 23(2), 81-89.
Bartlett, P. L., Baxter, J., & Weaver, L. (2011). Experiments with Infinite-Horizon, Policy-
Gradient Estimation (No. arXiv: 1106.0666).

34
Barzilay, R., & Elhadad, N. (2002). Inferring strategies for sentence ordering in
multidocument news summarization. Journal of Artificial Intelligence Research, 17,
35-55.
Benčić, F., Nemet, S., Barać, M., & Kukolj, D. (2018, November). Trends in AI based
Automotive Industry using Patent Analysis. In 26th TELFOR 2018.
Blockeel, H., Dehaspe, L., Demoen, B., Janssens, G., Ramon, J., & Vandecasteele, H. (2002).
Improving the efficiency of inductive logic programming through the use of query
packs. Journal of Artificial Intelligence Research, 16, 135-166.
Bod, R. (2002). A unified model of structural organization in language and music. Journal of
Artificial intelligence research, 17, 289-308.
Brafman, R. I., & Tennenholtz, M. (2003). Learning to coordinate efficiently: A model-based
approach. Journal of Artificial Intelligence Research, 19, 11-23.
Brookings Institution website. (2020). Wolff J. How to improve cybersecurity for artificial
intelligence, Brookings Institution’s Artificial Intelligence and Emerging Technology
(AIET), https://www.brookings.edu/research/how-to-improve-cybersecurity-for-
artificial-intelligence/ ; June 9th, 2020 [accessed July 14th, 2020].
Brooks, R. A. (1999). Cambrian intelligence: The early history of the new AI. MIT press.
Built In website. (2020). Schroer A., Artificial Intelligence in Cars Powers an AI Revolution in
the Auto Industry, https://builtin.com/artificial-intelligence/artificial-intelligence-
automotive-industry , Dec 19th, 2019 ; [accessed July 9th, 2020].
Cadoli, M., Donini, F. M., Liberatore, P., & Schaerf, M. (2000). Space efficiency of
propositional knowledge representation formalisms. Journal of Artificial Intelligence
Research, 13, 1-31.
Calvanese, D., Lenzerini, M., & Nardi, D. (1999). Unifying class-based representation
formalisms. Journal of Artificial Intelligence Research, 11, 199-240.
Cambridge dictionary website. (2020);
https://dictionary.cambridge.org/de/worterbuch/englisch/artificial-intelligence ;
[accessed Feb 3rd, 2020].
Chan, Y. K., Chen, Y. F., Pham, T., Chang, W., & Hsieh, M. Y. (2018). Artificial intelligence in
medical applications. Journal of healthcare engineering, 2018.
Charzinski J. (2019). Artificial Intelligence in Mobile Communication Networks, Hochschule
der Medien, http://maucher.pages.mi.hdm-stuttgart.de/ai/res/mobileNetworks.pdf.

35
Chen, J. C., Patel, V. M., & Chellappa, R. (2016). Unconstrained face verification using deep
cnn features. In March 2016 IEEE winter conference on applications of computer
vision (WACV) (pp. 1-9). IEEE.
Cloud Painter website. (2020). https://www.cloudpainter.com [accessed June 22th, 2020]
Cooper, M. J., Jain, P., & Sidhu, H. S. (2020). Systems and methods for training machine
models with augmented data .U.S. Patent Application No. 16/598,956.
Crevier, D. (1993). AI: the tumultuous history of the search for artificial intelligence. Basic
Books, Inc.
Cubillos, F. A., & Lima, E. L. (1998). Adaptive hybrid neural models for process control.
Computers & chemical engineering, 22, S989-S992.
Daily Sabah website. (2020). S. Güner. AI and avatars: Are you ready for your digital twin?,
https://www.dailysabah.com/life/ai-and-avatars-are-you-ready-for-your-digital-
twin/news, March 18th, 2020 ; [accessed Nov 2nd, 2020].
De Toni, A., Nassimbeni, G., & Tonchia, S. (1996). An artificial, intelligence‐based
production scheduler. Integrated Manufacturing Systems.
Der Bank blog website. (2020). Seidl P., Kurz erklärt: Künstliche Intelligenz in der
Finanzbranche, https://www.der-bank-blog.de/finanzbranche-kuenstliche-
intelligenz/technologie/37663753/ , May 6th, 2020 ; [accessed Nov 2nd, 2020].
Deutsche Welle. (2019). D. Pundy. Wenn der Richter mit Softwarehilfe urteilt,
https://www.dw.com/de/wenn-der-richter-mit-softwarehilfe-urteilt/a-47174579 ,
Dec 27th, 2019 ; [accessed Oct 3rd, 2020].
Ebay Press Room. (2018). Neue Bildersuche in eBay-App startet in Deutschland,
https://www.ebayinc.com/stories/press-room/de/neue-bildersuche-in-ebay-app-
startet-in-deutschland/ , press release, Nov 14th, 2018 ; [accessed Sept 3rd, 2020].
Electrek News website. (2020). Lambert F., Tesla’s acquisition of DeepScale starts to pay
off with new IP in machine learning, https://electrek.co/2020/04/17/tesla-
acquisition-deepscale-new-ip-machine-learning/ , April 17th, 2020 ; [accessed July
13th, 2020].
Elektrotechnik Automatisierung website. (2019). K. Juschkat. Wie Künstliche Intelligenz in
der Produktion helfen kann. https://www.elektrotechnik.vogel.de/wie-kuenstliche-
intelligenz-in-der-produktion-helfen-kann-a-803174/ , Feb 26th 2019 ; [accessed Nov
2nd, 2020].

36
Eliot, L. (2009). Pink brain, blue brain: How small differences grow into troublesome gaps--
and what we can do about it. Houghton Mifflin Harcourt.
Emerj website. (2019). Sennaar K., Machine Learning in Genomics – Current Efforts and
Future Applications https://emerj.com/ai-sector-overviews/machine-learning-in-
genomics-applications/ , Nov 28th, 2019 ; [accessed Nov 2nd, 2020].
Forbes website. (2020). Marr B. The Top 10 Artificial Intelligence Trends Everyone Should
Be Watching In 2020. https://www.forbes.com/sites/bernardmarr/2020/01/06/the-
top-10-artificial-intelligence-trends-everyone-should-be-watching-in-
2020/#2344d39d390b , Jan 6th, 2020 ; [accessed Nov 2nd, 2020].
Franklin, J. (2003). The representation of context: ideas from artificial intelligence. Law,
Probability and Risk, 2(3), 191-199.
Frantz, R. (2003). Herbert Simon. Artificial intelligence as a framework for understanding
intuition. Journal of Economic Psychology, 24(2), 265-277.
Gamberger, D., & Lavrac, N. (2002). Expert-guided subgroup discovery: Methodology and
application. Journal of Artificial Intelligence Research, 17, 501-527.
Global Market Insight Website. (2020). Wadhwani P., Kasnale S., Artificial Intelligence (AI)
in Automotive Market Size, https://www.gminsights.com/industry-analysis/artificial-
intelligence-ai-in-automotive-market , Dec 2019 ; [accessed July 10th, 2020].
Goldberg, D. E., & Holland, J. H. (1988). Genetic algorithms and machine learning.
Gordon, D. F. (2000). Asimovian adaptive agents. Journal of Artificial Intelligence Research,
13, 95-153.
Hassoun, M. H. (1995). Fundamentals of artificial neural networks. MIT press.
Heartbeat website. (2020). Cohen J., Computer Vision at Tesla,
https://heartbeat.fritz.ai/computer-vision-at-tesla-cd5e88074376 ; [accessed Nov
2nd, 2020].
Hopfield, J. J. (1982). Neural networks and physical systems with emergent collective
computational abilities. Proceedings of the national academy of sciences, 79(8), 2554-
2558.
IBM website. (2020). Artificial Intelligence in medicine. https://www.ibm.com/watson-
health/learn/artificial-intelligence-medicine ; [accessed Feb 12th, 2020].

37
Ingenieur.de website. (2020). Künstliche Intelligenz – versteckt in zahlreichen Geräten,
https://www.ingenieur.de/technik/fachbereiche/elektronik/kuenstliche-intelligenz-
versteckt-in-zahlreichen-geraeten/ , Dec. 30th, 2019 [accessed Feb 2nd 2020].
Jim Whitehurst. (2019). AN EXECUTIVE’S GUIDE TO REAL-WORLD AI Lessons from the Front
Lines of Business. Harvard Business Review Analytic Services,
https://enterprisersproject.com/sites/default/files/an_executives_guide_to_real_w
orld_ai.pdf.
Kaminka, G. A., Pynadath, D. V., & Tambe, M. (2002). Monitoring teams by overhearing: A
multi-agent plan-recognition approach. Journal of artificial intelligence research, 17,
83-135.
Kojić, N., Reljin, I., & Reljin, B. (2006). Neural network for optimization of routing in
communication networks. Facta universitatis-series: Electronics and Energetics,
19(2), 317-329.
Kusters, R., & Borgida, A. (2001). What's in an attribute? Consequences for the least
common subsumer. Journal of Artificial Intelligence Research, 14, 167-203.
Langley, P. (2011). The changing science of machine learning. 275-279.
Larsson, G., Maire, M., & Shakhnarovich, G. (2016, October). Learning representations for
automatic colorization. In European conference on computer vision (pp. 577-593).
Springer, Cham.
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. nature, 521(7553), 436-444.
Lerman, K., Minton, S. N., & Knoblock, C. A. (2003). Wrapper maintenance: A machine
learning approach. Journal of Artificial Intelligence Research, 18, 149-181.
Lionbridge website. (2020). Einsatzbeispiele für KI in der Finanzbranche,
https://www.lionbridge.com/de/blog/data-services/einsatzbeispiele-fuer-ki-in-der-
finanzbranche/ , April 16th, 2020 ; [accessed Nov 2nd, 2020].
Markram, H. (2006). The blue brain project. Nature Reviews Neuroscience, 7(2), 153-160.
McCarthy, J., Minsky, M. L., Rochester, N., & Shannon, C. E. (2006). A proposal for the
dartmouth summer research project on artificial intelligence, august 31, 1955. AI
magazine, 27(4), 12-12.
Meek, C. (2001). Finding a path is harder than finding a tree. Journal of Artificial Intelligence
Research, 15, 383-389.

38
Motor Zeitung website. (2019). In Berlin machen die Autos blitzschnell ihren Führerschein,
http://motorzeitung.de/news.php?newsid=547256 , Jan 15th, 2019 ; [accessed July
9th, 2020].
Muggleton, S. (2014). Alan Turing and the development of Artificial Intelligence. AI
communications, 27(1), 3-10.
Nature website. (2018). Webb S., Deep learning for biology,
https://www.nature.com/articles/d41586-018-02174-z , Feb 20th, 201 ; [accessed
July 9th, 2020].
Nguyen, T. T., Nguyen, C. M., Nguyen, D. T., Nguyen, D. T., & Nahavandi, S. (2019). Deep
learning for deepfakes creation and detection. arXiv preprint arXiv:1909.11573.
Nilsson, N. J. (2014). Principles of artificial intelligence. Morgan Kaufmann.
Nock, R. (2002). Inducing Interpretable Voting Classifiers without Trading Accuracy for
Simplicity: Theoretical Results, Approximation Algorithms. Journal of Artificial
Intelligence Research, 17, 137-170.
Oke, S. A. (2008). A literature review on artificial intelligence. International journal of
information and management sciences, 19(4), 535-570.
Otani, T., Toube, H., Kimura, T., & Furutani, M. (2017). Application of AI to mobile network
operation. ITU Journal: ICT Discoveries, Special Issue, 1, 1-7.
Parkes, D. C., & Wellman, M. P. (2015). Economic reasoning and artificial intelligence.
Science, 349(6245), 267-272.
Patel-Schneider, P. F., & Sebastiani, R. (2003). A new general method to generate random
modal formulae for testing decision procedures. Journal of Artificial Intelligence
Research, 18, 351-389.
Portugal, I., Alencar, P., & Cowan, D. (2018). The use of machine learning algorithms in
recommender systems: A systematic review. Expert Systems with Applications, 97,
205-227.
Power, Y., & Bahri, P. A. (2005). Integration techniques in intelligent operational
management: a review. Knowledge-Based Systems, 18(2-3), 89-97.
Railway Technology website. (2017). Facing fears: should facial recognition technology be
installed in railway stations?, https://www.railway-
technology.com/features/featurefacing-fears-should-facial-recognition-technology-
be-installed-in-railway-stations-5942494/ ; [accessed Sept 8th, 2020].

39
Reiter, O., Rotemberg, V., Kose, K., & Halpern, A. C. (2019). Artificial Intelligence in Skin
Cancer. Current Dermatology Reports, 8(3), 133-140.
Renzi, C., Leali, F., Cavazzuti, M., & Andrisano, A. O. (2014). A review on artificial intelligence
applications to the optimal design of dedicated and reconfigurable manufacturing
systems. The International Journal of Advanced Manufacturing Technology, 72(1-4),
403-418.
Rich, E. (1983). Artificial Intelligence. McGraw-Hi11.
Rosati, R. (1999). Reasoning about minimal belief and negation as failure. Journal of
Artificial Intelligence Research, 11, 277-300.
RSIP Vision website. (2020). Machine Vision For Robotics,
https://www.rsipvision.com/machine-vision-for-robotics/ ; [accessed July 13th,
2020].
Samsung Newsroom website. (2020). Samsung Electronics Unveils 2020 QLED 8K TV at CES.
Samsung press release, https://news.samsung.com/global/samsung-electronics-
unveils-2020-qled-8k-tv-at-ces, Jan 6th, 2020 ; [accessed July 10th, 2020].
Science Daily website. (2019). Center for Genomic Regulation, Artificial intelligence applied
to the genome identifies an unknown human ancestor.
https://www.sciencedaily.com/releases/2019/01/190116122650.htm, Jan 16th 2019;
[accessed July 10th, 2020].
Sendra, S., Rego, A., Lloret, J., Jimenez, J. M., & Romero, O. (2017, May). Including artificial
intelligence in a routing protocol using software defined networks. In 2017 IEEE
International Conference on Communications Workshops (ICC Workshops) (pp. 670-
674). IEEE.
Shabbir, J., & Anwer, T. (2018). Artificial intelligence and its role in near future. arXiv
preprint arXiv:1804.01396.
Sharmila, T., & Leo, L. M. (2016, April). Image upscaling based convolutional neural network
for better reconstruction quality. In 2016 International Conference on
Communication and Signal Processing (ICCSP) (pp. 0710-0714). IEEE.
Sivadasan, B. (2018). Application of Artificial Intelligence in Electrical Engineering. Global
Research and Development Journal for Engineering, National Conference on
Emerging Research Trend in Electrical and Electronics Engineering. 2455-5703.

40
Slideshare website. (2017). Joseph M. Artificial Intelligence In Power Systems,
https://de.slideshare.net/mathewjoseph2811/artificial-intelligence-in-power-
systems-seminar-presentation, ROLL NO: 21 REG. NO:14018761 , Sept 25th, 2017;
[accessed July 13th, 2020].
Soderland, S., & Lehnert, W. (1994). Wrap-up: a trainable discourse module for information
extraction. Journal of Artificial Intelligence Research, 2, 131-158.
SSI-SCHAEFER website. (2018). Fauland J., Artificial Intelligence in Logistics; White Paper SSI
Schaefer; https://www.ssi-
schaefer.com/resource/blob/566408/06d87a3eff1abfbdd7af3875404b724a/white-
paper-artificial-intelligence-in-logistics--dam-download-en-16558--data.pdf, Oct
2018; [accessed Nov 2nd, 2020].
Stay Relevant website. (2017). Artificial Intelligence As a Music Composer,
https://stayrelevant.globant.com/en/artificial-intelligence-composing-original-
music/ , Sept 19th, 2017; [accessed Oct 2rd, 2020].
SUI, D., & JIN, X. H. (2011). Network Congestion Control Method Based on Artificial
Intelligence. Computer Simulation, (9), 30.
Synced website. (2019). Li H. The Online Dating Industry Loves Artificial Intelligence,
https://syncedreview.com/2019/06/15/the-online-dating-industry-loves-artificial-
intelligence/ , June 15th, 2019 ; [accessed Sept 3rd, 2020].
Synced website. (2020). Lin Z., AI Offers Video Game Design Possibilities Far Beyond Virtual
Reality, https://syncedreview.com/2020/06/13/ai-offers-video-game-design-
possibilities-far-beyond-virtual-reality/ , June 13th, 2020; [accessed July 8th, 2020].
Tech News website. (2020). Scientists use artificial intelligence to fight the corona virus,
https://www.gsmgotech.com/2020/01/scientists-use-artificial-intelligence.html , Jan
2020; [accessed May 3rd, 2020].
Techiexpert website. (2019). Ocampos J., Artificial Intelligence 2020: Trends, Predictions in
the World of AI, https://www.techiexpert.com/artificial-intelligence-2020-trends-
predictions-in-the-world-of-ai/ , 10/2019; [accessed Nov 2nd, 2020].
Tennenholtz, M. (2002). Competitive safety analysis: Robust decision-making in multi-
agent systems. Journal of Artificial Intelligence Research, 17, 363-378.
Teslamag website. (2020). Teslas KI-Chef zeigt Videos mit geretteten Fußgängern und
erklärt Autopilot-Software, https://teslamag.de/news/teslas-ki-chef-videos-

41
 gerettete-fussgaenger-autopilot-software-erklaert-28107, April 23th, 2020; [accessed
June 24th, 2020].
The New Stack. (2020). Mok K., Deep Learning AI Generates Realistic Game Graphics by
Learning from Videos https://thenewstack.io/deep-learning-ai-generates-realistic-
game-graphics-by-learning-from-videos/ , Jan 10th, 2019 ; [accessed July 8th, 2020].
Tilghman, P. (2019). Will rule the airwaves: A DARPA grand challenge seeks autonomous
radios to manage the wireless spectrum. IEEE Spectrum, 56(6), 28-33.
Time website. (2019). Purtill C. Stop Me if You've Heard This One: A Robot and a Team of
Irish Scientists Walk Into a Senior Living Home , https://time.com/longform/senior-
care-robot/ , Oct 4th, 2019; [accessed July13th, 2020].
Time website. (2020). Coronavirus Researchers Are Using High-Tech Methods to Predict
Where the Virus Might Go Next, https://time.com/5780683/coronavirus-ai/ , Feb.
2020 ; [accessed May 3rd, 2020].
Tolosana, R., Vera-Rodriguez, R., Fierrez, J., Morales, A., & Ortega-Garcia, J. (2020).
Deepfakes and beyond: A survey of face manipulation and fake detection. arXiv
preprint arXiv:2001.00179.
Towards Data Science website. (2019). Kuprenko V., How AI Changes the Logistic Industry,
https://towardsdatascience.com/how-ai-changes-the-logistic-industry-
3d55401778d, Jul 3rd, 2019; [accessed Nov 2nd, 2020].
Tsai, P. F., Chu, J. Z., Jang, S. S., & Shieh, S. S. (2003). Developing a robust model predictive
control architecture through regional knowledge analysis of artificial neural
networks. Journal of Process Control, 13(5), 423-435.
Turney, P. D. (1994). Cost-sensitive classification: Empirical evaluation of a hybrid genetic
decision tree induction algorithm. Journal of artificial intelligence research, 2, 369-
409.
Uraikul, V., Chan, C. W., & Tontiwachwuthikul, P. (2007). Artificial intelligence for
monitoring and supervisory control of process systems. Engineering applications of
artificial intelligence, 20(2), 115-131.
Varela, F. J., Thompson, E., & Rosch, E. (2016). The embodied mind: Cognitive science and
human experience. MIT press.
VHR Company website, (2020). McKeon C., 6 Automotive Artificial Intelligence Trends to
Watch, https://blog.v-hr.com/blog/6-automotive-artificial-intelligence-trends-how-

42
 ai-is-already-changing-the-automotive-industry , June 21st, 2018 ; [accessed July 10th,
2020].
Vulture website. (2020). Lee C. ‘We Were All Very Nervous’: How The Irishman’s Special
Effects Team Got the Job Done. https://www.vulture.com/2020/01/how-the-
irishman-used-cgi-and-special-effects-on-actors.html, Vulture Magazine, Jan 9th
2020; [accessed July 3rd, 2020].
Walker, M. A. (2000). An application of reinforcement learning to dialogue strategy
selection in a spoken dialogue system for email. Journal of Artificial Intelligence
Research, 12, 387-416.
Walsh, W. E., & Wellman, M. P. (2003). Decentralized supply chain formation: A market
protocol and competitive equilibrium analysis. Journal of Artificial Intelligence
Research, 19, 513-567.
Wei, N., Li, C., Peng, X., Zeng, F., & Lu, X. (2019). Conventional models and artificial
intelligence-based models for energy consumption forecasting: A review. Journal of
Petroleum Science and Engineering, 181, 106187.
Wenzel, A., Linke, M., Meßmer, T., Micard, G., Schubert, G., Minde, A., & Kindl, M. (2019).
Innovative grid optimization approach based on artificial neural networks. In 2019
IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe) (pp. 1-4). IEEE.
Wikipedia website . (2020).
https://en.wikipedia.org/wiki/Artificial_intelligence_in_video_games ; [accessed July
8th, 2020].
Wong, A. K. C., & Wang, Y. (2003). Pattern discovery: a data driven approach to decision
support. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications
and Reviews), 33(1), 114-124.
World Economic Forum. (2019). Fleming S. The rise of the robo-restaurant: AI is ready to
take your order, https://www.weforum.org/agenda/2019/09/the-rise-of-the-robo-
restaurant/ , Sept 17th, 2019 ; [accessed July 13th, 2020].
Xu, G., Mu, Y., & Liu, J. (2017). Inclusion of artificial intelligence in communication networks
and services. ITU J., ICT Discoveries, Special, (1), 1-6.
Zanuttini, B. (2003). New polynomial classes for logic-based abduction. Journal of Artificial
Intelligence Research, 19, 1-10.

43
ZD Net website. (2020). Palmer D. AI is changing everything about cybersecurity, for better
and for worse. Here's what you need to know, https://www.zdnet.com/article/ai-is-
changing-everything-about-cybersecurity-for-better-and-for-worse-heres-what-you-
need-to-know/, March 2nd, 2020 ; [accessed July 14th, 2020].
Zhang, C., Patras, P., & Haddadi, H. (2019). Deep learning in mobile and wireless
networking: A survey. IEEE Communications Surveys & Tutorials, 21(3), 2224-2287.
Zhang, R., Isola, P., & Efros, A. A. (2016, October). Colorful image colorization. In European
conference on computer vision (pp. 649-666). Springer, Cham.
Zhong, G., Wang, L. N., Ling, X., & Dong, J. (2016). An overview on data representation
learning: From traditional feature learning to recent deep learning. The Journal of
Finance and Data Science, 2(4), 265-278.
Zou, J., Huss, M., Abid, A., Mohammadi, P., Torkamani, A., & Telenti, A. (2019). A primer on
deep learning in genomics. Nature genetics, 51(1), 12-18.

Prof. Dr.-Ing. Jens Christian Timmermann is professor at Baden-

Wuerttemberg Cooperative State University at the technical campus of
Friedrichshafen. He manages the B.Eng. study program “Communications
with focus on traffic systems” within Electrical Engineering.
Research Interests: RF Propagation, Channel Modeling, Mobile
Communications, Trends in (Electrical) Engineering
Address: Baden Wuerttemberg Cooperative State University (DHBW)
Ravensburg, Campus Friedrichshafen, Fallenbrunnen 2, 88045
Friedrichshafen, Germany. Tel. +49 7541 2077-263,
E- mail: [email protected]

Dr.-Ing. Atheer L. Salih Al-Tameemi is research staff member in the field of

Electrical Engineering at Baden-Wuerttemberg Cooperative State
University at the campus of Friedrichshafen. He works in different areas of
research within Electrical Engineering.
Research Interests: Simulation, Image Processing, Communication
Engineering, Wave Propagation, Control systems, EMC Systems.
Address: Baden-Wuerttemberg Cooperative State University (DHBW)
Ravensburg, Campus Friedrichshafen, Fallenbrunnen 2, 88045
Friedrichshafen, Germany. Tel. +49 7541 2077-227,
E- mail: [email protected]

44
Herausgeber

Prof. Dr. Heinz-Leo Dudek

Prorektor und Dekan der Fakultät für Technik

Duale Hochschule Baden-Württemberg Ravensburg

Baden-Wuerttemberg Cooperative State University
Marienplatz 2
88212 Ravensburg

ISBN 978-3-945557-09-9
ISSN 2199-238X
DOI 10.12903/ DHBW_RV_FN_03_2020_TIMMERMANN_AL-TAMEEMI

View publication stats