ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

Avinash Kumar Rajbhar

Artificial Intelligence And Data Science
Engineering Department,
A. C. Patil College of Engineering,
Kharghar, Navi Mumbai, Maharashtra
410210.

Abstract— The field of artificial intelligence (AI) has shown

an upward trend of growth in the 21st century (from 2000 to
2015). The evolution in AI has advanced the development of
human society in our own time, with dramatic revolutions
shaped by both theories and techniques. However, the
multidisciplinary and fast- growing features make AI a field in 2) ARTIFICIAL INTELLIGENCE
which it is difficult to be well understood.

Keywords— Artificial Intelligence; Machine learning;

Artificial intelligence is a core of computer science [6,
7]. It is a scientific development through which a
INTRODUCTION machine is developed to perform actions that fall within
Artificial Intelligence (AI) has grown dramatically the scope of human intelligence, such as teaching, logic,
and becomes more and more institutionalized in the self-correction, and self-programming. The principal aim
21st Century. In this era of interdisciplinary science, of artificial intelligence is to make computers do the
work that humans do [8]. Artificial intelligence is vital to
of computer science, cybernetics, automation,
some scientific and applied fields. Today, artificial
mathematical logic, and linguis- tics [1], questions
intelligence has pervaded people’s daily activities ,
have been raised about the specific concept of AI [2].
including industry, agriculture, and medicine. For
Actually, as early as the 1940s and 1950s, scien- tists example, physicians use a robot to perform surgeries
in the field of Mathematics, Engineering, and with high accuracy. The abilities that artificial
Computer Science had explored the possibilities of intelligence can advance are learning, pattern-
artificial brains and were trying to define the recognition, reasoning, problem solving, visual
intelligence of the machine. In 1950, Turing [3] perception, and language comprehension. Furthermore,
presented the famous ‘‘Turing Test’’ which defined artificial intelligence investigation is highly technical,
of the concept of ‘‘Machine Intelligence’’. On this specialized, and relies heavily on logic, knowledge,
back- ground, the origins of AI can be traced to the planning, education, communication, imaging, and the
workshop held on the campus of Dartmouth College ability to move and handle objects. This science is
in 1965 [4], in which McCarthy persuaded divided into two types:
participants to accept the concept of ‘‘Artificial
A) Applied: It is the more common type that includes
Intelligence’’. It is likewise the beginning of the first
intelligent systems capable of accomplishing one
‘‘Golden age’’ of AI.
task, such as driving a car without a driver. This
In simple terms, AI aims to extend and augment type is termed weak or narrow artificial intelligence.
the capac- ity and efficiency of mankind in tasks of
B) Generalized: This is the less common type and
remaking nature and governing the society through involves systems that can handle any assignment
intelligent machines, with the final goal of realizing a because they have sufficient capacity to solve any
society where people and machines coexist obstacles or unusual situations. This type is called
harmoniously together [5]. robust. To date, no example of robust artificial
1) THE SCIENCES intelligence is available, because it is in a stage of
This section reviews the gains in artificial intelligence, evolution and growth.
machine learning, and deep learning. The information
presented is valuable especially for those interested in 3) MACHINE LEARNING
these sciences. Figure 1 illustrates the main differences
between the three sciences With the technological developments in recent years, new
terms have begun to emerge. Big data [9] [10], Industry
4.0 [11], and artificial intelligence are the most popular
ones. Although machine learning is not as popular as these
terms, it is a concept that has been on the rise. However, ,
many questions surround machine learning. Machine
learning is applied in different sectors and applications today,
and its use is increasing gradually. Machine learning is a sub-
area of artificial intelligence [12, 13]. Information
technology systems automatically learn patterns and
relationships from data and gain without being explicitly

Fig 1.Differences between the three sciences

programmed. Machine learning has been successfully CONCLUSION
supported in business, investigation, and improvement for This article explores the profound impact of Artificial
many years [14–17]. Furthermore, machine learning can Intelligence (AI) and Machine Learning (ML) in
automatically produce knowledge, train algorithms, identify contemporary industries, highlighting their boundless
relationships, and recognize unknown patterns. These potential. In the current age of big data, AI and ML play
identified patterns and relationships can be utilized to a new, pivotal roles in healthcare, utilizing sophisticated algorithms
unknown data set in order to make predictions and optimize for data analysis, while grappling with challenges such as
processes. Machine learning algorithms can be categorized data scarcity and extensive computational demands..
into three main divisions:
• Decision Process: Machine learning techniques are REFERENCES
utilized to create predictions or classifications of specific [1] D. Vernon, G. Metta, and G. Sandini, ‘‘A survey of artificial
data, which can be labelled or unlabeled. The outcome of cognitive systems: Implications for the autonomous development of
mental capabil- ities in computational agents,’’ IEEE Trans. Evol.
these techniques is to estimate a pattern in the data, which Comput., vol. 11, no. 2,pp. 151–180, Apr. 2007.
assists in decision making. [2] D. Kirsh, ‘‘Thinking with external representations,’’ Ai Soc., vol. 25,
• Error Function: This function judges a model’s no. 4,pp. 441–454, 2010.
prediction and whether it gives actual or false effects. [3] A. M. Turing, ‘‘Computing machinery and intelligence,’’ Mind, vol.
59, no. 236, pp. 433–460, Oct. 1950.
Additionally, this function has the ability to make a
[4] P. McCorduck, Machines Who Think. Natick, MA, USA: A. K.
comparison to judge the functioning of the model and verify Peters, 2004.
the functioning of machine learning techniques.
[5] T. J. M. Bench-Capon and P. E. Dunne, ‘‘Argumentation in artificial
• Model Optimization Process: If the proposed intelligence,’’ Artif. Intell., vol. 171, nos. 10–15, pp. 619–641, 2007.
model for implementation fits well with the training data set, [6] N. Miailhe and C. Hodes, “The Third Age of Artificial Intelligence,”
Open Edition Journal, Special Issue, vol. 17, pp. 6–11, 2017.
then the weights will be adjusted to reduce the discrepancy
between the proposed work and the model estimation. The [7] A. N. Ramesh, C. Kambhampati, J. R. T. Monson, and P. J. Drew,
“Artificial intelligence in medicine,” The Annals of The Royal
technology will repeat this assessment, improve the College of Surgeons of England, vol. 86, pp. 334–338, 2004.
implementation process, and improve the weights [8] C. M. Signorelli, “Can Computers Become Conscious and Overcome
independently until the accuracy is met. Humans?,” Frontiers in Robotics and AI, vol. 5, pp. 1–20, 2018.
• Supervised learning: In general, this type includes [9] M. G. Kibria, K. Nguyen, G. P. Vilardi, O. Zhao, K. Ishizu, and F.
Kojima, “Big Data Analytics and Artificial Intelligence in Next-
most of the problems in machine learning, which is Generation Wireless Networks,” Arxiv, pp. 1–10, 2018.
characterised by looking at training samples. Each sample is [10] A. H. Ali, M. Z. Abdullah, S. N. A. Abdul-Wahab, and A. M, “A
entered as X so that it corresponds to a specific result, which Brief Review of Big Data Analytics Based on Machine Learning,”
is y. We need to train a model (mathematically is x→y Iraqi Journal for Computer Science and Mathematics, vol. 1, no. 2,
relationship mapping f) in unknown samples x after giving, pp. 13–15, 2020.
then we can obtain y predictions. If the prediction is a [11] D. Mhlanga, “Artificial Intelligence in the Industry 4.0, and Its Impact
on Poverty, Innovation, Infrastructure Development, and the
discrete value (often category types, such as spam/snail mail Sustainable Development Goals: Lessons from Emerging
in the mail classification problem, such as whether a user Economies?,” Sustainability, vol. 13, no. 5788, pp. 1–16.
will/will not buy a particular product), then it will be termed [12] N. Kühl, M. Goutier, R. Hirt, and G. Satzger, “Machine Learning in
a classification problem. If the prediction result is a Artificial Intelligence: Towards a Common Understanding,” in
continuous value (e.g., apartment prices, stock prices, etc.), Proceedings of International Conference on DSystem Sciences
then this state will be termed a regression problem. (HICSS-52), pp. 1–11, 2019.
[13] M. M. Mijwil, “Malware Detection in Android OS Using Machine
• Unsupervised learning: It is a form of learning in Learning Techniques,” Data Science and Applications (DataSCI), vol.
which information is categorized or not. Unsupervised 3, no. 2, pp. 5–9, 2020.
learning finds hidden patterns in data. It uses them to infer [14] M. M. Mijwil and I. E. Salem, “Credit Caqird Fraud Detection in
from datasets entered into the system, without labelled data. Payment Using Machine Learning Classifiers,” Asian Journal of
Computer and Information Systems, vol. 8, no. 4, pp. 50–53, 2020.
Given that no classification has previously been done, the
[15] K. Kersting, “Machine Learning and Artificial Intelligence: Two
system can classify using data sets. Fellow Travelers on the Quest for Intelligent Behavior in Machines,”
• Reinforcement learning: It is a form of learning Frontiers in Big Data, vol. 1, pp. 1–4, 2018.
that takes place between supervised learning and [16] M. M. Mijwil, I. E. Salem, and A. R. A, “Utilisation of Machine
Learning Techniques in Testing and Training of Different Medical
unsupervised learning. It is used for the same applications as Datasets,”
supervised learning. Large amounts of unlabelled data and [17] S. N. Abd, M. Alsajri, and H. R. Ibraheem, “Rao-SVM Machine
small amounts of labelled data are commonly used. Learning Algorithm for Intrusion Detection System,” Iraqi Journal for
Computer Science and Mathematics, vol. 1, no. 1, pp. 23–27, 2020.

Fig 2.Categories of machine learning