COM6905 Research Methods And Professional Issues.docx

COM6905 Research Methods And Professional Issues
Answer:
Introduction
Recent years have seen a huge increase in data and analytics capabilities as a result of the
fast exponential rise of data and the development of increasingly complex algorithms.
Computational power has risen in lockstep with the growth in storage capacity. Future
technology will impact enterprises as a result of these rapid technical breakthroughs. Ultra
intelligence is a distinguishing characteristic of the X 4.0 era. In this article, we will discuss
machine learning within the framework of artificial intelligence and present a succinct
summary of the X 4.0 era.
Big data is defined as datasets that contain the following characteristics: (1) heterogeneous
and autonomous sources, (2) diverse dimensions, (3) sizes and/or formats that defy
conventional processes or tools for effectively and affordably capturing, storing, managing,
analyzing, and exploiting; and (4) complex, dynamic, and evolving relationships
acknowledge that organizations are increasingly challenged with big data difficulties and
that a varied range of technologies for accumulating, manipulating, organizing, analyzing,
and displaying them should be developed and used [1]. Current big data strategies, which
incorporate parts of statistics, applied mathematics, and computer science, are inadequate,
and enterprises seeking benefit from big data must adopt more adaptive, trustworthy, and
interdisciplinary approaches.
Businesses are repurposing big data as a beneficial resource. It is generated in a multitude
of ways, including through the internet, sensors, mobile phones, payment systems, cameras,
telematics, and wearable devices. Its worth becomes apparent as it gets more extensively
utilized. "As data becomes increasingly commoditized, value is expected to flow to owners
of unusual data, actors that combine data in creative ways, and, most importantly,
producers of good analytics," they write. Data and analytics are reshaping the competitive
environment. Leading firms are harnessing their strengths to develop totally new business
models while also improving their basic operations. The network effects of digital platforms
have produced a winner-take-all dynamic in some businesses."
Numerous disruptive methods are based on big data and analytics. Massive data integration

capabilities have the potential to disrupt institutional and technical silos by delivering novel
insights and analytical tools, as well as novel data perspectives such as orthogonality [2].
Electronic Communication Networks (ECNs), for example, are enormously scalable E-
commerce platforms capable of instantly linking customers and sellers, transforming
inefficient marketplaces. Granular data may be used to tailor products and services (for
example, as part of Industry 4.0) – and, perhaps most intriguingly, health care. Innovative
analytic techniques have the potential to significantly accelerate innovation and discovery.
Above all, data and analytics can help you make better and more timely decisions.
Numerous sectors are already undergoing upheaval as a result of big data and analytics, and
a new wave of disruption is on the horizon as automated learning advances, endowing
robots with incredible thinking, decision-making, and communication skills. In this
research, we describe a reinforcement learning framework based on the GOWDA system
that is capable of intelligently de-noising signals via wavelet transformation while
maintaining information.
The X 4.0 Era: Evolutionary Aspect
Through the combination of cyber-physical systems (such as the Internet of Things),
information and communication technology (ICT), and cloud computing, Industry 4.0
ushers in a new era of data sharing and production automation. The phrase "Industries 4.0"
refers to the fourth industrial revolution [2]. With the introduction of Internet technology, it
is commonly regarded as the application of the generic notion of cyber-physical systems to
industrial production. Similar concepts have been introduced in the United States by
General Electric and in China by the State Council, respectively, under the banners of
Industrial Internet and Made in China 2025.
Three hypotheses have been underlined in order to fully comprehend the notion of cyber-
physical systems: "(1) Manufacturing systems' communication infrastructure will become
more cost effective, enabling wider use. It has a purpose. Only a few examples include the
engineering, configuration, servicing, diagnosis, operation, and maintenance of goods, field
equipment, machinery, and plants. It will cement its position as a critical component of
future industrial systems. (2) Field gadgets, machinery, plants, and factories (as well as
individual goods) will become more networked (e.g., the Internet or a private factory
network).
They do this through the establishment of a virtual live presence on the internet, replete
with unique identities. They will be used to store information such as documents, three-
dimensional (3-D) models, simulation models, and other types of data. This content is
updated on a regular basis and so reflects the most recent version. Along with the data,
numerous functionalities will be applied to real things, such as negotiating, exploration, and
so on. These data objects complement the physical equipment with which they are attached
and provide a second identity on the network, serving as a knowledge base for a variety of

applications. "The originality of this scenario is not in the introduction of fresh technology,"
they write. "Rather, it is in the novel combination of existing technologies." The availability
of large amounts of data opens up a slew of new possibilities. DaaS, like other "as a service"
(aaS) models, is predicated on the notion that the product (in this case, data) may be
delivered to the user on demand regardless of the provider's geographic or organizational
distance from the consumer. Additionally, the rise of service-oriented architecture (SOA)
has rendered irrelevant the physical platform on which data is kept.
Tim Burners-Lee, creator of the World Wide Web and one of Time Magazine's "100 Most
Influential People of the Twenty-First Century," introduced the notion in 1989 [1]. The
internet and associated technologies have changed substantially during the last two
decades. Web 1.0 was a network focused on cognition, but web 2.0 was a network based on
expression. Since the web's creation, four generations have emerged: web 2.0 as a medium
for communication, web 3.0 as a medium for association, plus web 4.0 as a media for
incorporation. The focus of Web 4.0 is on the "hyper-intelligent electronic agent."
Web 1.0 was initially intended to serve as a platform for individuals and organizations to
exchange broadcast information. The early web allowed for limited user engagement and
content creation, limiting users to little more than searching for and reading information.
File and web servers, content and business portals, search engines, personal information
managers, e-mail, peer-to-peer file sharing, and publish and subscribe technologies were all
created during this time period.
The word "Web 2.0" was devised in 2004 by Dale Dougherty, founder and CEO of Maker
Media, Inc. He coined the term "read-write web." At this level, web 2.0 technologies include
blogs, wikis (such as Wikipedia), social bookmarking, social networking sites (such as
Facebook and MySpace), instant messaging, mash-ups, and auction websites (such as eBay)
(e.g., Linked-in). The Web 3.0 platform is comprised of two major components: semantic
technology and social computing. Ontologies, semantic search, glossaries and classifications,
peculiar intellectual digital aides, and information bases are only a few of the essential
technologies now being studied.
Once Web 3.0 technologies such as improved natural language processing are firmly
established on the internet, the capacity to construct intelligent systems capable of thinking
(such as learning and reasoning) emerges as an emergent property. As a result of enabling a
mutually beneficial relationship between humans and machines, Web 4.0 is also referred to
as the symbiotic web. With web 4.0, it will be possible to create more intelligent interfaces
in which machines collect data and respond by executing and prioritizing tasks.
Business Intelligence And Analysis
The importance of business intelligence and analytics (BI&A) has grown as a result of the
amount and severity of data-related challenges confronting today's organizations. BI&A 1.0

systems are mostly based on 1970s statistical methodologies and 1980s data mining
techniques. The era of Web 3.0 (mobile and sensor-based) has begun with the advent of
mobile interfaces, visualization, and human-computer interaction design. The convergence
of the physical and virtual worlds in BI&A 4.0 has resulted in multichannel strategies that
encompass online, offline, and online-to-offline interactions. Machine learning employs an
inductive technique to develop a model of the world from the data it receives. It is capable
of updating and improving its representation in response to fresh data.
Deep neural networks with several hidden layers are utilized in this field of machine
learning. The feedforward and recursive neural networks are two of the most frequently
utilized forms of deep neural networks [4]. Convolutional neural networks are widely used
to recognize pictures through the processing of a hierarchy of characteristics — for
example, linking a nose to a face and finally to a complete cat. This capability of picture
recognition is critical for the development of autonomous cars, which must constantly
detect their surroundings. On the other hand, recursive neural networks are utilized when
the complete sequence and context are critical, like in speech recognition and natural
language processing.
Reinforcement Learning
Reinforcement learning, on the other hand, drives behavior toward a stated objective, i.e.,
the value functions are codified. The algorithms test a range of different actions before
agreeing on the most successful ones, which includes a creative aspect.
This collection of techniques employs multiple machine learning methods to obtain more
accurate predictions than any single method could achieve alone, resulting in ensemble
methods, which employ multiple learning algorithms to obtain more accurate predictions
than any of the constituent learning algorithms could achieve alone.
Assume that the observational equation for X is as follows:
Xt = S(t) + Nt , t ∈ T = {1, . . . , n(= 2J )}
where n is the aggregate number of recurrently appraised time facts, S(t) signifies the
unidentified function that denotes the signal at time t, and Nt denotes the preservative noise
variables distributed independently and identically and experimented at time t.
Reinforcement Learning
The objective of reinforcement learning (RL) is to teach an agent how to formulate and
behave optimally in a given circumstance, where the optimum policy is the least expensive.
When an agent is in state s, the value function V(s) indicates the efficacy, or predicted cost,
of the policy. It may also be expressed recursively as Equation (2) or in terms of the Bellman

equation as Equation (3), where the value of equals the immediate cost of state transfer plus
the values of the potential following states weighted by the transition probability and a
discount factor γ.
V π (s) = E{ X∞ i=0 γ i ct+i}
= E{ct + γV π (st+1)|s = st}
= X s 0 T(s, π(s), s0 )(C(s, a, s0 ) + γV π (st+1))
The best policy π ∗ with the minimum cost V π ∗ , satisifies V π ∗ (s) ≤ V π (s), ∀s ∈ S and ∀a
∈ A. V ∗ (s) = arg min a 0 X s 0 T(s, π(s), s0 )(C(s, a, s0 ) + γV π (st+1)).
There are two main types of reinforcement learning techniques (see?). The first technique
does not require a model, but the second method does. Following a series of investigations
and changes, the agent will directly generate the best policy utilizing model-free
methodologies. Model-based methods will construct a model from the obtained data and
then utilize the constructed model to identify the ideal approach.
Simulation Study
The simulation research is conducted to determine the enactment of the anticipated
method. This simulation research accomplishes two objectives. To begin, we demonstrate
that the new strategy outperforms the standard method for each signal.
Statistical data
We use Monte Carlo simulations to produce mistakes (jumps) from two separate patterns in
order to illustrate (1) extreme volatility (Pattern I) and (2) excessive volatility with Markov-
switching multifractals (Pattern II). For each pattern under consideration, we build a time
series data collection with a total of 29 samples. A sine function with an equal amplitude
and frequency distribution is used to determine the trend. Following the simulation
employed by, we add jumps to this trend in order to create Pattern I signals. The magnitude
of the leap is normally distributed with a mean of zero and a unit variance of zero, and the
occurrences of jumps are uniformly distributed (with a Poisson arrival rate).
Future Work
The better performance of the GOWDA reinforcement learning algorithm enables
automated analytics and helps consumers to upsurge the productivity of their big data-
driven policymaking. The Design I data demonstrate a distinctive stylized fact about data,
namely, heavy tails or excessive fluctuation, whereas the Pattern II data demonstrate
excessive fluctuation with Markov-switching multifractals.

FinTech is an industry comprised of enterprises that leverage existing resources to compete
in the market for financial services provided by traditional financial institutions and
intermediaries. FinTech is a buzzword for new financial services applications, procedures,
products, and business models.
Figure 1: Panel 1 illustrates excessive volatility, whereas Panel 2 employs Markov-switching
multifractals.
References
A. B. D. &. K. T. Novak, "Product decision-making information systems, real-time sensor
networks, and artificial intelligence-driven big data analytics in sustainable Industry 4.0. ,"
Economics, Management and Financial Markets, pp. 16(2), 62-72, 2021.
S. &. M. J. Cohen, "Cyber-Physical Process Monitoring Systems, Real-Time Big Data Analytics,
and Industrial Artificial Intelligence in Sustainable Smart Manufacturing. ," Cyber-Physical
Process Monitoring Systems, Real-Time Big Data Analytics, and Industrial Artificial
Intelligence in Sustainable Smart Manufacturing. , p. 16(13), 2021.
C. &. C. Y. Zhang, "A review of research relevant to the emerging industry trends: Industry
4.0, IoT, blockchain, and business analytics. ," Journal of Industrial Integration and
Management, , pp. 5(101), 165-180, 2020.
S. P. J. H. C. G. S. &. D. Y. K. Bag, "Role of institutional pressures and resources in the adoption
of big data analytics powered artificial intelligence, sustainable manufacturing practices and
circular economy capabilities," Technological Forecasting and Social Change, pp. 163,
120420, 2021.
E. Grant, "Big Data-driven Innovation, Deep Learning-assisted Smart Process Planning, and
Product Decision-Making Information Systems in Sustainable Industry," Economics,
Management, and Financial Markets,, pp. 16(1), 9-19., 2021.

COM6905 Research Methods And Professional Issues.docx

More Related Content

Similar to COM6905 Research Methods And Professional Issues.docx (20)

More from write31 (20)

Recently uploaded (20)

COM6905 Research Methods And Professional Issues.docx