INTRODUCTION technology is still not that wide and great.

It can
be sometimes costly. Second, is that the model
Artificial intelligence (AI) and deep learning and development standards of the AI technology
have been one of the most rising technologies to is still developing and is yet to be fully realized
date. It has been applied in different types of and be matured enough. Lastly, when a new
fields and industries. This review will talk about technology arrives in the world and is to be
2 applications of these technology specifically applied in many industries, there will always be
from the healthcare system: dentistry, and the a question of practicality and if it does or does
network security systems on intrusion detection not violate rules, laws and any code of ethics in
systems. the industry.

HEATHCARE APPLICATION As mentioned in [1], the medicine industry has

used diagnostic imaging being central in many
Shown in [1], is an application of artificial healthcare fields, Digital health data are
intelligence in dentistry. It particularly reviews ubiquitously collected, AI allows the integration
the chances and challenges of applying artificial of different and heterogenous data domains, it
intelligence, machine learning, and deep facilitates research and discovery by adding in
learning in the dental industry such as the silico experimentation options to conventional
possibilities, limitations of AI-based dental research hierarchies, complementing other
diagnostics, treatment planning, and conduct, for research levels and existing modeling strategies
example, image analysis, prediction making, and lastly, AI may streamline routine work and
record keeping, as well as dental research and increase the face-to-face time doctors/dentists
discovery. and their patients have. In dentistry,
convolutional neutral networks have been
The problem with AI solutions in these field is adopted in research settings from 2015 onwards,
that it is not largely been practiced due to the mainly on dental radiograph.
mentioned reasons of “1) limited data
availability, accessibility, structure, and In summary, AI application on healthcare
comprehensiveness, 2) lacking methodological systems would largely be helpful in the future.
rigor and standards in their development, 3) and For example, in [1], it is mentioned that data
practical questions around the value and sources can be federated in the future as opposed
usefulness of these solutions, but also ethics and to being centralized thereby having data from
responsibility.” For the first reason given in the multiple institutions. Today, we have supervised
article, it is indeed that the accessibility of the learning used in the medical field but a

1
probability of having an unsupervised or semi- The proposed model uses multi-layered of
supervised learning in the future. recurrent neural networks designed to be
implemented for Fog computing security that is
The article shows us the existing situation of very close to the end-users and IoT devices. The
important applications of Artificial Intelligence model is composed of two major engines: traffic
and how it can still be improved. AI technology analysis engine and classification engine. The
is in fact not yet matured but when it will be, traffic connection records are pre-processed in
there are possible revolutionary changes that traffic processing unit leading to traffic data in a
will not only change the medical/dental fields format suitable to be processed by the deep
but also other fields as well. neural network of classification engine whereas
the connections are classified into normal and
INTRUTION DETECTION SYSTEM attack by intelligent intrusion detection engine.
The model adopts a recurrent neural network
In [2], the topic is about the deep recurrent trained by an adaptive version of
neural network for Internet of Things (IoT) backpropagation algorithm for enhanced
intrusion detection system. It is universally prediction capability of the normal/attack
known that as digital technology is further classification. The demonstration of the model
developed more problems will occur. But was done using a balanced version of the
despite the problems, more people continue to challenging dataset: NSL-KDD. The
develop it since problems naturally occur when performance of the model was measured using a
developing something and we are left to deal variety of typical metrics with added additional
with the problems as part of the developing metrics of Mathew correlation and Cohen’s
process. Similarly, as the internet becomes more Kappa coefficients for deeper insights.
deeply integrated into industrial operations and
platforms, more possible dangers may occur. Results in [2] shows that the effectiveness of the
Such is the case of intrusions, cybercrimes, and adaptive cascaded filtering using the recursive
other related problems. Many security systems structure of neural networks where each network
and models have been proposed and/or adaptively tuned to different
implemented and one of the premier choices is parameters/hyperparameters for enhancing the
intrusion detection systems using artificial detection of specific intrusion types. This
intelligence. The paper presents an artificially resulted to the model showing high sensitivity to
full-automated intrusion detection system for DoS attacks that represent one of the prominent
Fog security against cyber-attacks. attacks thwart the development of IoT network
besides detecting other types of attacks’

2
categories such as Probe, R2L, and U2R in a
competitive computation overhead as each
record required 66 micro seconds on average to
be processed. It was concluded that the proposed
model is capable of properly and efficiently
working in real time environments.

REFERENCES

[1] F. Schwendicke, W. Samek and J. Krois, "Artificial Intelligence in Dentistry: Chances and
Challenges," Journal of Dental Research, vol. 99, no. 7, p. 769–774, 2020.

[2] M. Almiani, A. AbuGhazleh, A. Al-Rahayfeh, S. Atiewi and A. Razaque, "Deep recurrent neural
network for IoT intrusion detection system," Simulation Modelling Practice and Theory, vol. 101,
no. 102031, 2020.