Assignment On

Review Questions About Artificial Intelligence and Machine Learning

Course: Information and Communication Technology: Japan and Bangladesh
Course code: JS 402

Submitted to
Dr. Md. Ahsan Habib
Associate Professor, Electrical and Electronics Department
Faculty of Engineering Technology, University of Dhaka

Submitted by
Tazriba Akter Akhi
Roll: KM-197-014
2nd year 4th semester, 4th batch
Department of Japanese Studies, University of Dhaka
 Date of Submission: 18 May 2023

1. Define machine learning and describe the main types of machine learning algorithms.

Machine learning is a branch or subset of Artificial Intelligence and computer science. It is one of
the exciting field if artificial intelligence and important component of the growing field of data
science. Machine learning focuses on the use of data and algorithms to intimate the way that
humans learn, gradually improving its accuracy.
The term “Machine Learning” is coined by the American Computer Scientist “Arthur Samuel” in
1959.
Today machine learning is used around the world in social media features, virtual assistance, image
recognition and so on.

Types of machine learning :

Machine learning models fall into three primary categories. The three main machine learning
algorithms types are:
1. Supervised machine learning
The most common type of learning is supervised learning, which works similarly to utilizing
flash cards in the classroom to train a model to predict future outputs based on past input and
output data.
2. Unsupervised machine learning
Unsupervised Learning is really different. The algorithm is fed a data set and given the means
to grasp its qualities rather than having a preferred result preselected
3. Reinforcement machine learning
Similar to the classical conditioning technique used by Ivan Pavlov to make dogs drool when
they hear a bell, reinforcement learning trains a computer through reinforcement mechanisms.
Early on, a reinforcement learning system will produce a lot of errors. Over time, though, the
machine picks up on signals indicating accuracy or errors and improves by learning from them.
 2. What is Artificial intelligence, and how it is different from machine learning?

The field of computer science known as artificial intelligence is focused on developing computers
that can mimic human intelligence and problem-solving skills. In order to streamline and advance
in the future, they accomplish this by ingesting a vast amount of data, processing it, and learning
from the past while a typical computer software would require human intervention to rectify errors
and enhance workflow.
Artificial intelligence and machine learning are sometimes used interchangeably. But there are
slight differences between AI and ML.
SL Artificial intelligence Machine learning
no
1. AI is a board field ML is a subset of AI
2. AI is used to build machines and ML is the application of AI that automatically enables a
computers that have the ability to machine or system to learn and improve from
mimic cognitive functions experience.
associated with human intelligence,
such as being able to understand,
and respond to spoken or written
language, analyse data, make
recommendations, and more.
3. AI allows a machine to simulate ML allows a machine to learn autonomously from past
human intelligence to solve data
problems.
4. AI results in knowledge or making ML results in data.
the system intelligent.
5. The goal of AI is to develop an The goal of ML is to build machines that can learn from
intelligent system that can perform data to increase the accuracy of the output.
complex tasks
6. AI has wide scope to applications ML has limited scope of application.
7. AI is used in healthcare, finance, ML is used for optical charter recognition, web security,
transportation, aviation, marketing, imitation learning etc.
media, education etc.
 3. Explain the concept of Neural networks, and describe how they are used in machine
learning?

An artificial intelligence technique called a neural network instructs computers to analyse data in
a manner modelled after the human brain. Deep learning is a sort of machine learning that employs
interconnected neurons or nodes in a layered framework to mimic the human brain. It develops an
adaptive system that computers utilize to continuously learn from their errors and improve.
Artificial neural networks so make an effort to more accurately tackle complex issues, such as
summarizing documents or identifying faces.
Computers can make intelligent decisions with minimal human intervention thanks to neural
networks. This is why they can learn and model complex, nonlinear relationships between input
and output data.

Artificial neurons are coupled in three layers of a simple neural network:

• Input Layer
The input layer is where data from the outside world enters an artificial neural network.
• Hidden Layer
It is possible for artificial neural networks to have a lot of hidden layers. The input layer or other
hidden layers serve as the input for hidden layers.
• Output Layer
There may be one or more nodes in the output layer. It displays the end result of every data
processing operation carried out by an artificial neural network.
Neural networks in machine learning :

A group of methods known as neural networks are used in machine learning to enable computers
to recognize patterns without explicit programming.
Machine learning uses neural networks, which process data using mathematical or computational
models.
Due to the non-linear nature of these neural networks, they are typically able to simulate
complicated interactions between data input and output and identify patterns in datasets.

Neural network applications in machine learning typically fall into one of these three broad
categories:
• A neural network can classify data by spotting patterns and sequences.
• Analysis using regression and functional approximation
• Clustering and data filtering are all types of data processing.
There are several benefits to machine learning with neural networks, including:

• They maintain information across the whole network, enabling the neural network to
continue operating even if some of the information is lost in one area.
• Once neural networks have been trained on a good data set, they can evaluate and present
data more quickly, saving both time and money. Additionally, they are less likely to make
mistakes, particularly if they were trained using high-quality data.
• Results from neural networks are accurate and of high quality.
• There are several benefits to machine learning with neural networks, including:
• Once neural networks are trained on a good data set, they save money and time because
they evaluate data more quickly.
 4. What are some common applications of Machine Learning in industries such as
healthcare, finance and retail?

The science of machine learning is expanding quickly and has countless potential uses. Machine
learning will alter numerous sectors over the coming years, including manufacturing, retail, and
healthcare.
Machine learning can be applied to manufacturing for quality assurance, automation, and
customisation. Machine learning, for instance, can be used to find product flaws before they are
sold to customers. Additionally, it can be utilized to automate repeated processes like those on a
manufacturing line. And finally, to tailor items to specific customers, firms will increasingly apply
machine learning.
Machine learning can be used in retail to analyze data and assist firms in making better decisions
regarding inventory and pricing. Retailers will increasingly use machine learning to offer products
to customers based on their prior behavior. This is known as personalization. Robotics will also be
employed more frequently, and machine learning will be used to automate processes like order
pickup and shelf stocking.
Machine learning can be applied to healthcare for prevention, treatment, and diagnosis. For
instance, machine learning can be used to make earlier and more precise disease diagnoses.
Personalized treatments based on a patient's traits can also be created using it. Additionally,
machine learning can be utilized to identify illness risk factors and make personalized suggestions
for leading healthy lives.
We have only begun to scratch the surface of machine learning's potential thus far. We will witness
even more incredible uses of this revolutionary technology as it develops.
In the upcoming years, machine learning is expected to revolutionize a variety of industries.
Machine learning in retail will make it possible to analyze data more precisely, personalize goods
and services, and even utilize robots in-store. Machine learning will completely change
healthcare's approach to diagnosis, treatment, and prevention. Additionally, machine learning in
manufacturing will enhance quality assurance, automate procedures, and enable more
customization. These are just a few instances of how machine learning will alter the way that
business is done as we currently know it. It's time to start preparing for the machine learning
revolution, no matter what industry you're in.
 5. How does unsupervised learning differ from supervised learning, and what are some
examples of each?

Both supervised and unsupervised learning are considered machine learning techniques, and which
one you choose will rely on your dataset's structure, size, and application scenarios.
Both methodologies are applied to various situations and datasets. Below is a table outlining the
similarities and differences between the two learning approaches.
Supervised learning algorithms are trained using labelled data and unsupervised learning
algorithms are trained using unlabelled data.
The supervised learning model uses direct feedback to determine whether or not it is foretelling
the correct result. On the other hand, a model of unsupervised learning does not incorporate
feedback.
The results are predicted by the supervised learning model while unsupervised learning models
find the hidden patterns data.
In supervised learning, the model receives input data in addition to output. In unsupervised
learning, the model receives only input data.
The objective of supervised learning is to develop the model's capacity for output prediction in the
presence of novel data. On the contrary, unsupervised learning aims to extract hidden patterns and
insightful information from an unknown dataset.
To train the model, supervised learning requires supervision. The model can be trained without
any supervision using unsupervised learning.
A supervised learning model yields reliable results. In comparison to supervised learning, an
unsupervised learning model could produce less accurate results.
Supervised learning falls short of actual artificial intelligence since we must first train the model
for each set of data before it can make accurate predictions. On the other hand, unsupervised
learning is more in line with actual artificial intelligence because it acquires knowledge by
experience, much like a kid does when learning daily tasks.

Examples:
Supervised learning comprises a variety of procedures, including Bayesian logic, decision trees,
support vector machines, multi-class classification, linear regression, and logistic regression.
Unsupervised learning contains a number of algorithms, including the Apriori algorithm, KNN,
and Clustering.
 6. What are the some of ethical concerns associated with the use of AI in decision making
and automation, and how can they be addressed?

Situations that arise as a result of a moral dilemma and require resolution are referred to as ethical
issues. Thus, moral concerns frequently conflict with the foundational values of a community.
Significant ethical difficulties and questions are raised by the quick development and acceptance
of AI in decision-making processes. The potential for bias and discrimination is one of the key
ethical consequences of AI in decision-making. Because AI systems are only as good as the data
they are trained on, biased or discriminating data will result in biased or discriminatory results
from he AI system. Existing socioeconomic inequities may be sustained and even amplified by
this.
The absence of transparency and accountability in AI decision-making is another ethical
implication. It can be tricky to grasp how judgments are being made and whether they are fair and
just because AI systems can be sophisticated and difficult to comprehend. Due to this, judgments
made with AI may not be held accountable or responsible for their results.
Concerns exist on how AI will affect employment and the workforce as well. As AI systems
progress, they may eventually replace human labour, which could result in job loss and economic
inequity.
It is crucial to take into account these moral ramifications and create laws and policies that
guarantee the ethical development and application of AI in decision-making. This entails
encouraging openness, responsibility, and justice in AI systems and making sure they adhere to
moral standards and ideals.
To adequately deal with the ethical concerns surrounding AI, we should adhere to some guidelines.
First, a governance board-style ethics council committee should be established to handle concerns
related to risk and issues with fairness, privacy, cyber, and other data.
Reducing ethical difficulties requires developing an ethical AI framework. Increasing public
understanding can aid in lowering ethical concerns with AI.
Overall, by adhering to ethical AI principles including justice, social being, privacy, protection and
security, dependability and safety, and value of alignment, ethical AI concerns related to the usage
of AI can be reduced.
 7. Discuss the history of AI research and development in Japan, and identify some of the
key players and innovations in the field.

In terms of robots and artificial intelligence (AI), Japan has historically been the international
leader. Japan produced $5.7 billion worth of robots in 2000. They produced and transported an
enormous quantity of robotics in 2021 alone, accounting for close to 50% of the global market
share.
But normally the question arise that how did they rise to the top of the robotics world, and how
has it affected Japanese culture.
The Japanese government views artificial intelligence (AI) as one of the key components that will
contribute to "Society 5.0," a future civilization in which online and physical space are more
closely intertwined.
Japan is improving worker safety, lowering the chance of error, and enabling other more
established technologies to advance at a faster rate through committed investment in robotic
technologies. They are leading the robotics revolution in an effort to boost production and
efficiency for their society and economy.
Currently, the main applications of robots in Japan are: Production, Science and Medicine
Hospitality.
Operational robots are currently the most prevalent type in Japan and are mostly used in the
industrial sector. They can be found at factories all around the nation, especially those that produce
electronics and vehicles. Around a quarter of a million industrial robot workers are "employed" in
the nation to complete activities like welding, painting, putting parts together, and carrying heavy
objects.
Tokyo's museums, hotels, and cafés annually draw thousands of tourists and residents because of
the diverse ways that robots are used outside of workplaces.
The study has highlighted application areas like those related to marketing, information services,
and security that could present further prospects for European businesses, in addition to the AI and
robotics sector and automated driving technologies.
Japanese businesses offer marketing-related AI-applied solutions in the areas of data analysis,
automated marketing, and picture recognition. Additional economic opportunities may exist in
these and other fields where Europe has expertise.
 8. What are some of the challenges and opportunities facing AI in Japan today, and how
might they impact the country's economy and society in the future?

Japan's artificial intelligence sector has mostly concentrated on robotics as one of its sub-domains,
developing AI particularly in the fields of technology for industrial and medical uses. Artificial
intelligence is a system that processes different types of data to produce predictions, suggestions,
and judgments using both machine intelligence and human thinking abilities.

According to the Nomura Research Institute, artificial intelligence will experience exponential
growth in Japan by 2035, when robots using AI may do half of all professions there. However,
while the AI market in Japan has been concentrating on robotics, abroad companies have focused
more on software development, providing a market entry opportunity for foreign enterprises.

Challenges:

Manufacturing companies in Japan's AI sector are currently faced with the difficulty of figuring
out how to put into practice the idea of creating robots that are specifically tailored to meet the
needs of consumers. Aside from continuing to develop AI targeted at the service industries, such
as healthcare and long-term care, the latest developments in the Japanese AI industry are moving
toward the creation of self-driving cars, drones, smart cities, and factories in the upcoming years.

Studies from research institutions like AIST (the National Institute of Advanced Industrial Science
and Technology) and RIKEN, Japan's largest comprehensive research institution, where various
projects relating to Artificial Intelligence in Japan are routinely carried out, have contributed to
Japan's technological advancement. Automation, prediction, and perception are just a few of the
industries that AI and deep learning are revolutionizing. Due to its traditional business practices
for management, recruiting, and promotion, Japan is trailing behind the US and China in the use
of AI.
In a survey performed in Japan in March 2022, nearly 49% of participants said that the artificial
intelligence (AI) installed at their businesses was not being used to its full potential. It was the
most prevalent problem with relation to AI operation, followed by a shortage of staff who could
use it and a lack of adequate integration with current systems.
Opportunities:

In terms of hardware development, including the creation of robots and vehicles, Japan's AI Sector
has been in the forefront. However, when it comes to software creation, American and European
businesses continue to dominate.

People's priorities and interests in regard to AI have shifted as a result of Covid-19. For instance,
because people tended to value face-to-face engagement in business, AI that augments human
labor was traditionally restricted to certain channels. The preference today is for technical
approaches that minimize physical contact. As a result, now is a wonderful time to think about
cooperation between Japanese and European businesses, using their own strengths. Deep learning
algorithms created by European AI firms could benefit the Japanese video game and automobile
industries by offering software solutions that have not yet been investigated in the Japanese AI
market.

Software platforms are also increasingly becoming open-sourced because multi-party partnerships
can explore and use AI more effectively than doing it on their own. This would open up more
potential for collaboration between European and Japanese AI firms in fields including marketing,
information services, image recognition, automated driving technologies, and security.
For these reasons, European businesses creating AI technologies might offer Japan appealing
solutions in these areas. Creating alliances with regional industry leaders can be a step in the right
direction.

9. Explain the concept of deep learning and describe some of the most significant
breakthroughs and applications of the technology in recent years.

A technique used in artificial intelligence (AI) called deep learning teaches computers to interpret
data in a manner modelled after the human brain. Deep learning models can identify intricate
patterns in images, text, audio, and other types of data to generate precise analyses and forecasts.
Deep learning techniques can be used to automate processes that ordinarily require human
intellect, such as text-to-sound transcription or the description of photographs.
The goal of teaching computers how to think and learn like people, artificial intelligence (AI) aims
to achieve. Deep learning powers several applications of AI present in everyday items, such as the
following: digital assistants, fraud detection, voice-activated television remote controls, and
automatic facial recognition. Modern technology like self-driving cars, virtual reality, and others
depend on it greatly.
Using an algorithm or a predetermined set of steps, data scientists train computer files called deep
learning models to perform tasks. Businesses employ a range of deep learning technologies to
assess data and produce predictions.
There are several applications for deep learning in the sectors of automotive, aerospace,
manufacturing, electronics, and other industries. Deep learning has recently found widespread use.
Here are a few contemporary instances of deep learning:
• Deep learning models are used by self-driving automobiles to automatically recognize
people and road signs.
• Deep learning is used by defence systems to automatically mark regions of interest in
satellite pictures.
• Deep learning is used in medical image analysis to automatically identify cancer cells for
medical diagnosis.
• Deep learning techniques are used in factories to automatically identify when individuals
or objects are too close to machinery.

Compared to AI and ML, Deep Learning is a relatively young idea or technology. However, it has
just as much, if not more, potential to transform businesses and organizations in the years to come.

We are confident that Deep Learning will significantly alter our lives in a variety of ways, from
more sophisticated and lifelike personal assistants to fully autonomous autos.
10. Discuss the role of government policies and public-private partnerships in promoting AI
research and development in Japan.

The Japanese government has set a new research and development strategy, with a focus on using
artificial intelligence and big data to solve some of the country’s biggest problems.
The strategy involves building partnerships between government, industry and academia, to
develop science and technology to tackle challenges such as ageing, low productivity growth and
improving wellbeing.
Japan is open to collaboration and cocreation to establish this new advanced society,
notwithstanding its historical internal orientation. Investment seems to be in order given the
predicted expansion of the AI market. All of these findings suggest toward a real window of
opportunity for startups, academics, investors, or AI-based companies. Japan might be a
fascinating partner to advance your AI developments, depending on the precise skills required. For
instance, the Japanese government and a number of industry sectors are driving interest in
information regarding software development and product enhancement. It is important to
comprehend Japanese culture, spend time establishing trust, and promote productive cooperation
between Dutch and Japanese efforts if one wants to be a fascinating partner.
Japan is looking for foreign collaboration and collaborations more and more. The EU, and
subsequently the Netherlands, benefit from Japan sharing similar social principles for AI, which
provides a win-win edge over working with nations that do not share similar human-centric values.
Strategic actions like alliances, mergers, or collaborations can improve Japan and the Netherlands'
competitiveness. Instead of developing brand-new hardware items, businesses are becoming more
interested in upgrading existing products through software development. Japan's software
restrictions may be a barrier to some partnerships, but they might also present a chance to augment
with Dutch knowledge or another complimentary relationship.
Scholarships offered by Japan to international exchange students are an additional chance for them
to enhance their education. which can encourage collaborations.
Our daily lives and enterprises are changing significantly as a result of artificial intelligence (AI).
While AI offers revolutionary solutions to social issues, it also raises a number of questions
regarding privacy, security, fairness, and even democracy due to its unexpected nature,
inexplicability, and reflection or amplification of data biases. In response, in the late 2010s, a
number of countries, international organizations, and research institutions all around the world
started to publish a list of principles for human-centric AI.
Digital ministers are anticipated to debate the human-centric approach to AI during the 2023 G7
meeting in Japan, which may involve regulatory or non-regulatory policy tools.
Recently bill revising Japan's "Copyright Act" was approved by the National Diet, and the reform's
main goals were to give innovators the much-needed flexibility and legal security. The primary
goal of this modification was to remove any misunderstanding regarding the use of copyrighted
works for comprehension and analysis in order to promote innovative digital and Artificial
Intelligence (AI) services that are currently or will be available in the future.

In other words, Japan has made sure that AI cannot be hampered by copyright. On September 12,
the European Parliament will review and vote once more on its own copyright regulations. It is
important to consider how Japanese legislators correctly linked machine learning technologies
with the advancement of AI and balanced their copyright regulations to support their technological
goals.