Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
HOME FIREFIGHTING SYSTEM BASED ON MACHINE LEARNING AND
INTERNET OF THINGS
ALİ HUSSEİN ABDULQADER AL-ZANGANAH
Electrical and Computer Engineering dept., Altinbas University, Istanbul – Turkiye.
Email: [Link]@[Link], ORCID: 0009-0007-2198-7421,
Dr. HASAN ABDULKADER
Computer Engineering dept., Altinbas University, Istanbul – Turkiye.
Email: [Link]@[Link], ORCID 0000-5297-040X
Abstract
The smart home is now of interest to many researchers. The network of things has enabled homes to
become interactive away from the traditional homes that have lasted for a long time. The Web is an
important part of the various exchanges of current life. The Web of Things has expanded rapidly to include
many aspects of life today, including home appliances, alarm systems, and distributed storage systems for
information created by home appliances. The various innovations and parts make up a smart home, and
they all produce valuable information that can be used to guess home conditions and outdoor traffic. A
wonderful and comprehensive smart home fire suppression system based on IoT and machine learning is
presented with plan and framework for optimization. The frame controls and directs the security of the
house and its current conditions through the use of a group of sensors, where there is a unit that controls
and examines electrical tools and natural variables, and to obtain a protected and accurate frame, the idea
of artificial intelligence was used to determine the state of the flame. This work shows that AI computations
can advance the usefulness of a home computing framework and improve home security.
Keywords: Internet of Things (IoT), Smart Home, Fire system, Machine Learning.
1. INTRODUCTION
An essential part of new data innovation and an extension of web applications is Internet
of Things (IoT). The main ideas of the IoT include the automation of tasks and the
connection of ordinary objects to the Web [1]. Accessory tools obtain information about
the actual world, which is then manipulated, analyzed, and used to carry out tasks. The
IoT applications in a wide number of companies, including boardroom establishment,
smart homes, cool transportation, and medical services [2].
Monitoring home security and climate is essential due to the high rate of electrical
accidents, individual hazards to home occupants, and damage to property. The security
of people's lives and property is of paramount importance and should be the primary
concern. Then the home should include a smart framework for checking from a distance,
entitlement and illuminating the occupant regarding exercises. A great home automation
framework is expected to bring home security, luxury, comfort and control. A savvy house
is a (IoT) application that permits its occupants to just screen, control, and manage their
home tasks from anyplace. The IoT, as per [3], is an organized framework that empowers
electronic items to talk with each other and trade information.
By checking and dealing with the home climate, savvy home mechanization frameworks
are essential to guaranteeing a decent personal satisfaction. The primary objectives of a
April 2023 | 143
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
savvy home mechanization framework are energy and water investment funds, wellbeing
(like the recognition of risky gases, fires, or house intruders), and comfort (through remote
observing and the board of machines and the actual climate).
The IoT home equipment advances of sensors and actuators for correspondence and
mechanization are connected with machines and contraptions, giving restricted or remote
home control [3], making the home smart by offering types of assistance that require
minimal human info or association. One has some control over a home, for example,
through the Web from any area, utilizing particular programming or a portable application
that sudden spikes in demand for a PC, tablet, PC, or cell phone [4].
Through the arranged IoT home equipment innovations of sensors and actuators for
correspondence and computerization, offering confined or remote home control [5],
machines and gadgets are associated with offer types of assistance that expect
practically zero human information or collaboration. This makes the home smart by
offering these types of assistance. For example, utilizing the Web, an individual could
deal with their home from anyplace, with the control component being given by particular
programming or a portable application that sudden spikes in demand for a PC, tablet, PC,
or cell phone [5].
The utilization of AI calculations in brilliant home computerization has the advantage of
dissecting and evaluating information to conjecture ecological activities and conditions
and further develop robotization [6]. A shrewd home robotization framework performs
staggeringly well when AI is coordinated with IoT advances in the plan and improvement
process. This article will focus on the smart home, which gives knowledge, control and
network to the average home. It is a component of the IoT annotation era in general [7].
This post will analyze security and protection concerns despite the advent of the Internet
of Things and great homes.
2. RELATED WORKS
The improvement of ecological control, energy the board, home security, and different
areas of savvy home robotization frameworks is a shared objective of examination articles
and contemporary writing. In the space of the IoT, AI calculations have additionally been
utilized for arrangement, expectation, and examination. The region remembers articles
for the Web of Things that cover computerized brilliant homes and the utilization of AI to
canny frameworks.
Utilizing IoT innovation, Govindraj et al. [8] introduced a savvy home mechanization
framework to play the job of the conventional home robotization framework. A satellite
station and a radio recurrence handset are utilized in the proposed framework to oversee
and screen machines, temperature, movement, and gases in the home climate.
Rani et al. [9] proposed a characteristic language handling (NLP) and computerized
reasoning (man-made intelligence)- based voice-controlled home computerization
framework. Voice orders are sent over a cell phone and interpreted utilizing a preset
normal language handling medium to control domestic devices. The framework was not
April 2023 | 144
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
extended to incorporate different areas of home mechanization, like the control,
observing, and recognition of natural circumstances, gatecrashers, developments, and
so forth. The framework was exclusively used to work home devices.
A client-server-based strategy for savvy home mechanization was proposed by Gladence
et al. [10]. The proposed framework lays out association between the frameworks and
individuals utilizing AI methods and NLP thoughts. The client guides to play out specific
errands, like working entryways and home apparatuses and tuning in for vocal bed
development.
Mehmood et al. [11] .'s improvement of an item ID instrument for the control of savvy
home machines used a model view regulator engineering, haze of things, and item
recognition calculation as the establishment for the robotization framework. Through the
message lining telemetry transport (MQTT) mode, the Web of Things gadgets connected
with home apparatuses. This study showed how joining object distinguishing proof
strategies with profound learning calculations further develops object identification in a
brilliant home setting.
Khan et al. [12] introduced a strategy for controlling domestic devices in light of
examination and canny direction, utilizing support vector machines for security of IoT
gadgets and blockchain innovation for smart navigation. Machines may now be worked
remotely on account of an Android application. While going with choices on home devices
and their situations with, creators utilized a straight part.
A system for overseeing home machines like lights, fans, radiators, and outlets was
advanced by Taiwo et al. [3]. With the assistance of Bluetooth and Zigbee advancements
for correspondence and an Android versatile application for sending orders, the proposed
system empowers on-and off-site home control. The proposed Bluetooth innovation,
notwithstanding, restricted the +e correspondence range.
A home computerization framework was depicted by Garg et al. [13] for detecting and
keeping up with appropriate encompassing house conditions, as well as to give security
from robbery or peril utilizing sensors and home apparatus the executives. It recovers
orders that work the house through the Web utilizing a cloud data set. Time sensitive
boundaries from different home sensors are likewise put away in the cloud data set.
3. SMART HOME BASED ON IoT
The smart home (SH) is a vital IoT application that helps control and monitor homes [14].
Figure (1) shows the popularity patterns of the expression “cool home” and “network of
things” starting from 2013. According to Figure (1), it is clear that smart home and network
of things are on the rise. Cool Homes uses a few IoT-based innovations to offer a range
of utility instances that include controlling lights, home appliances, phone remote control,
and more [15].
April 2023 | 145
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
Figure 1: Functions of a smart house [16]
Rather than traditional abodes, the shrewd home consolidates standard private elements
as well as abilities for home development, organizing, brilliant apparatuses, and hardware
computerization. Furthermore, broad commitment prospects are offered, and,
surprisingly, related uses for different energy admissions are diminished [17].
Table 1: Features of Smart Home [16]
Smart home Control of draperies, sound and video hardware,
Devices lighting frameworks, computerized theater
frameworks, network machines, cooling, and so on.
Features of Control of domestic devices, lighting, phones, indoor
Smart Homes and outside controllers, thief cautions, natural
observing, infrared sending, and customized time
controls, in addition to other things.
Concerning Table (1), the idea of SH has a couple of key parts, some of which are
momentarily depicted here. One is the SH's fundamental framework for home
robotization. It alludes to coordinating or overseeing electronic home machines, like
lighting, security, sound and video, PC equipment, warming, and cooling frameworks [18].
A focal handling unit is customized to gather information from electrical and electronic
gear and afterward, in the wake of following different methodology, convey specific data
to different gadgets. Cell phones, controllers, PCs, and different points of interaction can
be generally used to control the gadgets. One more fundamental part of the SH is the
home organization. A stage for family data associates laptops, home electrical machines,
security frameworks, lighting frameworks, and WAN in shrewd homes, performs hardware
the board, and offers information and mixed media [19].
4. MACHINE LEARNING ALGORITHMS
Artificial intelligence has gotten new improvement chances alongside the speedy
progression of science and innovation. The practical characteristics of man-made
brainpower are reinforced by the joining of multidisciplinary hypothetical information into
machine innovation that depends on PC innovation, like measurements and calculation
intricacy. It is feasible to work on the materialness of AI calculations and give more
April 2023 | 146
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
comfort for the monetary improvement of the business by leading a reasonable
investigation of AI calculations and giving guidance reference for later AI improvement
[20].
4.1 Random Forest (RF)
Leo Breiman made the Random Forest [21], which is an assortment of unpruned grouping
or relapse trees made from an arbitrary example of preparing information. The elements
picked during the acceptance system are aimlessly. The expectations of the gathering
are consolidated (greater part vote in favor of grouping, normal for relapse). Each tree is
developed as per [22]:
On the off chance that the preparation set contains N cases, yet with substitution,
by haphazardly choosing N cases. The preparation set for developing the tree will
be this example.
The variable m is picked with the end goal that for M info factors, m<<M is provided
at every hub, m factors are picked aimlessly from the M, and the best parted on
these m is utilized to part the hub. The amount m is held consistent during the
development of the RF.
Each tree is developed to its full potential. Pruning is not done.
Generally speaking, Random Forest performs significantly better than single tree
classifiers like C4.5. Its generalisation error rate compares favourably to Adaboost's,
although it is more noise-resistant [22].
4.2 Light Gradient Boosting Classifier
In 2016, Microsoft MSRA fostered the Light gradient boosting machine (LightGBM), a
speedy and viable inclination helped choice tree (GBDT) strategy with an open-source
work reason. This approach upholds viable equal preparation and is used for arranging,
order, relapse, and numerous other AI strategy tasks. The LightGBM strategy, rather than
Xtream Inclination Supported (XGBoost), utilizes a histogram to stimulate preparing, save
memory, and apply a reasonable development plan with profundity limits [23].
The crucial idea of LightGBM is to discretize the congruity of drifting point eigenvalues
into k canisters and produce a histogram with a width of k. LigthGBM can store 8-bit whole
numbers, decreasing memory utilization to 1/8 of the first, and doesn't need an enormous
measure of pre-arranged results capacity. This uncertain segment no affects the
exactness of the LightGBM mode [24].
4.3 Hist Gradient Boosting Classifier
Inclination supporting purposes a method called hist slope helping relapse to prepare
choice trees all the more rapidly. Preparing trees that are added to a group can be added
substantially more rapidly by binning or discretizing. Subsequently, the information factors
are utilized to carry out the hist angle helping technique's calculation. Each tree that is
added to a troupe really tries to redress the anticipated missteps utilizing the gathering's
prior models [25].
April 2023 | 147
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
Alongside different methodologies, the hGBR strategy is utilized. For example, the scikit-
learn AI bundle, which gives a trial variant of the inclination helping procedure that upholds
the histogram approach, may be utilized to make it. It offers the Hist Gradient Boosting
Classifier and Hist Gradient Boosting Regressor classes specifically. The default GBR
execution given by the library is significant degrees more slow than the hGBR execution,
as indicated by the scikit-learn documentation [26].
5. SMART HOME SIMULATION
The smart home system was simulated based on the Cisco Payment Tracer program,
where a virtual server was set up to be used as an Internet service provider, and as it is
known, these servers are connected to main routers and then distributed to homes so
that the Internet is used inside the house as shown in the Figure (2).
With regard to the devices of the supposed house, the house contained a system of solar
cells for energy supply, and among them the house contained a number of devices and
electrical controls, including that the doors were programmed with RFID keys, as well as
lighting systems, air conditioning and cooling systems, garden watering system, and
finally the fire safety system.
This work deepened in controlling the fire system to increase the security of the residents
inside the house, as artificial intelligence and machine learning techniques were used to
control and control that system. The mechanism of the system's work revolved around
several axes, as these axes were linked to the environmental conditions inside the house,
as it relied on the internal external temperature for one of the rooms, for example, as the
internal temperature exceeded the external one by a large difference, which means that
there is an abnormal thermal glow in the room, which indicates By burning, in addition to
that, the amount of lighting indicates the glow, and your lie is the decrease in the amount
of moisture, all of which are factors indicating the presence of fires.
The framework of the fire system is related to all of the above, in addition to that, a sensor
has been placed to measure the level of carbon dioxide inside the room, as it is known
that combustion generates carbon dioxide gas, and from here it can be said that in the
event of a meeting of all these factors It is certain that a fire will occur inside a particular
room. With regard to machine learning and control of the system, a condition has been
set that if the amount of all of the factors mentioned changes in a certain direction, which
is the direction that indicates the presence of fire, the fire system will open its sprinklers
automatically for the purpose of extinguishing that fire.
On the other hand, only the rain sensor was placed to measure the level of the amount
of spraying of the system and whether it is working well or not, and in the event that the
system is working, the amount of humidity, temperature, and glow is continued to be read,
as the decrease in heat and the excess of humidity gives an indication that the fire is
extinguished, and this in turn instructs the system to close its sprinklers.
April 2023 | 148
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
Figure 2. IoT Home system simulation
6. HARDWARE DESCRIPTION AND DATASET COLLECTION
The Hardware description of the data collection circuit consist of the elements shown in
figure (3).
April 2023 | 149
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
Figure 3: Dataset collection circuit
LM35 Sensor: Two pieces of this sensor were connected to read the temperature
inside and outside the room in which the fire system was activated.
DHT11 Sensor: Two pieces of this sensor were connected to read the humidity level
inside and outside the room in which the fire system was activated.
SPG30 Sensor: This sensor was used to measure the level of carbon dioxide in the
internal and external environment, as carbon dioxide increases in the atmosphere
when there are fires.
Rain Sensor: With regard to the rain sensor, it has been employed to see if the fire
system is working well or not, in other words, it gives an indication of opening the fire
system in the event of a fire.
LDR: The photo resistance sensor, which is a variable resistance whose value
decreases when the light increases, which gives an indication of the presence of a
fire, as the light resulting from the fires reduces the value of that sensor
The sample of data set collected shown in Table (1), It contained eight characteristics
(internal temperature, external temperature, the amount of light intensity inside the room,
the amount of light intensity outside the room, the amount of internal humidity, the amount
of external humidity, the amount of dioxide inside the room, the amount of carbon dioxide
outside the room) and then the target Who decides whether the fire system will operate
or not.
April 2023 | 150
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
Table 2: Dataset sample
In In Out Out In Out In Co2 Out Open
Temp Humidity Temp Humidity Light Light Co2 Fire
21.9 94 20.8 66 22 21 0.85 0.77 1
21.9 95 20.7 69 23 20 0.87 0.75 1
21.9 94 20.8 66 24 21 0.76 0.69 1
22 93 21.0 67 25 19 0.82 0.59 1
22 94 21.5 64 26 20 0.86 0.65 1
21.7 95 20.0 74 27 20 0.76 0.65 1
21.7 96 20.1 68 25 21 0.81 0.66 1
7. RESULTS AND DISCUSSION
As mentioned and illustrated in Figure 4, performance is evaluated using the Confusion
Matrix (CM), or the matrix used to describe characteristic outcomes in classification
problems. In binary classification tasks, the incorrect choice receives a category 0 rating,
while the correct one receives a category 1 rating [27].
TN FP
FN TP
Figure 4: Confusion Matrix
The TP (1, 1) and TN (0, 0) scores in this case indicate that normal and pathological
consumption patterns can be accurately distinguished. Similar to the FP (0, 1) and FN (1,
0) scores, these indicate an incorrect classification of the instances with normal and
abnormal circumstances. And although FN observations comprise dishonest spending
patterns that were accurately predicted, FP accounts for those observations in CM that
were honest but unfairly forecasted. The model's performance is evaluated using CM in
terms of a number of parameters, including precision, accuracy, recall, and F1 score.
𝑇𝑃+𝑇𝑁
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 = 𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁 (1)
𝑇𝑃
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = 𝑇𝑃+𝐹𝑃 (2)
𝑇𝑃
𝑅𝑒𝑐𝑎𝑙𝑙 = 𝑇𝑃+𝐹𝑁 (3)
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛×𝑅𝑒𝑐𝑎𝑙𝑙
𝐹1 − 𝑆𝑐𝑜𝑟𝑒 = 2 × 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛+𝑅𝑒𝑐𝑎𝑙𝑙 (4)
April 2023 | 151
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
The Confusion Matrix Results get from the used algorithm shown in Figures (5, 6, 7)
respectively, when the measured values of the metrics shown in Table (3, 4, 5)
respectively.
Figure (5) Light Figure (6) Random Forest Figure (7) Hist Gradient
gradient boosting CM CM Boosting Classifier CM
Table 3: Light Gradient Boosting Results
Precision Recall F1-Score Support Accuracy
1.00 1.00 1.00 5827
99.85
1.00 1.00 1.00 4883
Table 4: Random Forest Results
Precision Recall F1-Score Support Accuracy
1.00 1.00 1.00 5827
99.93
1.00 1.00 1.00 4883
Table 5: Hist Gradient Boosting Classifier Results
Precision Recall F1-Score Support Accuracy
1.00 1.00 1.00 5827
99.71
1.00 1.00 1.00 4883
As shown in Tables above the maximum accuracy geted when the Random Forest
algorithm used that equal to (99.93%).
8. CONCLUSION
In smart homes, enhancing home security is a top difficulty, and solving this issue requires
constant technical advancement. A brand-new technology was put forth that would allow
the fire system to be managed according to the standards of heat, humidity, light intensity,
and carbon dioxide gas in the immediate vicinity. The IoT-based engineering upholds
constant observing of an individual's home condition as well as controller of domestic
devices across an organization. It additionally incorporates security caution, controller,
April 2023 | 152
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
distant natural observing, and different elements. The sensors and microcontroller have
been effectively associated with the cloud server. The data set effectively stores the
information. The aftereffects of all testing settings and perceptions exhibit the fulfillment
of this savvy home framework. The client can screen the state and circumstance at home
and approach information from mechanical devices. The establishment of this framework
will increase living expectations by improving security. The sensors and microcontroller
have been effectively associated with the cloud server. The data set effectively stores the
information. The aftereffects of all testing settings and perceptions exhibit the fulfillment
of this savvy home framework. The client can screen the state and circumstance at home
and approach information from mechanical devices. The establishment of this framework
will increase living expectations by improving security.
Acknowledgment
The researcher extends his thanks and appreciation to the Turkish University of Altinbaş for the support
and guidance it provided to him to complete this work.
References
1. M. Burhan, R. A. Rehman, B. Khan, and B. S. Kim, “IoT elements, layered architectures and security
issues: A comprehensive survey,” Sensors (Switzerland), vol. 18, no. 9, pp. 1–37, 2018, doi:
10.3390/s18092796.
2. V. Chang, Z. Wang, Q. Xu, L. Golightly, B. Liu, and M. Arami, “Smart Home based on Internet of Things
and Ethical Issues,” Sci. Technol. Publ., no. Femib, pp. 57–64, 2021, doi: 10.5220/0010178100570064.
3. O. Taiwo and A. E. Ezugwu, “Internet of Things-Based Intelligent Smart Home Control System,” Secur.
Commun. Networks, vol. 2021, 2021, doi: 10.1155/2021/9928254.
4. L. Oliveira, V. Mitchell, and A. May, “Smart home technology—comparing householder expectations at
the point of installation with experiences 1 year later,” Pers. Ubiquitous Comput., vol. 24, no. 5, pp.
613–626, 2020, doi: 10.1007/s00779-019-01302-4.
5. B. K. Sovacool, D. D. Furszyfer Del Rio, and S. Griffiths, “Policy mixes for more sustainable smart home
technologies,” Environ. Res. Lett., vol. 16, no. 5, 2021, doi: 10.1088/1748-9326/abe90a.
6. K. C. Yao, W. T. Huang, C. C. Wu, and T. Y. Chen, “Establishing an AI Model on Data Sensing and
Prediction for Smart Home Environment Control Based on LabVIEW,” Math. Probl. Eng., vol. 2021,
2021, doi: 10.1155/2021/7572818.
7. Y. I. Alzoubi, A. Al-Ahmad, and A. Jaradat, “Fog computing security and privacy issues, open
challenges, and blockchain solution: An overview,” Int. J. Electr. Comput. Eng., vol. 11, no. 6, pp. 5081–
5088, 2021, doi: 10.11591/ijece.v11i6.pp5081-5088.
8. V. Govindraj, M. Sathiyanarayanan, and B. Abubakar, “Customary homes to smart homes using
Internet of Things (IoT) and mobile application,” Proc. 2017 Int. Conf. Smart Technol. Smart Nation,
SmartTechCon 2017, no. April 2019, pp. 1059–1063, 2018, doi:
10.1109/SmartTechCon.2017.8358532.
9. P. J. Rani, J. Bakthakumar, B. P. Kumaar, U. P. Kumaar, and S. Kumar, “Voice controlled home
automation system using natural language processing (NLP) and internet of things (IoT),” ICONSTEM
2017 - Proc. 3rd IEEE Int. Conf. Sci. Technol. Eng. Manag., vol. 2018-Janua, no. 40, pp. 368–373,
2017, doi: 10.1109/ICONSTEM.2017.8261311.
10. L. M. Gladence, V. M. Anu, R. Rathna, and E. Brumancia, “Recommender system for home automation
using IoT and artificial intelligence,” J. Ambient Intell. Humaniz. Comput., no. 0123456789, 2020, doi:
April 2023 | 153
Tianjin Daxue Xuebao (Ziran Kexue yu Gongcheng Jishu Ban)/
Journal of Tianjin University Science and Technology
ISSN (Online):0493-2137
E-Publication: Online Open Access
Vol: 56 Issue: 04: 2023
DOI10.17605/[Link]/DGVQZ
10.1007/s12652-020-01968-2.
11. F. Mehmood, I. Ullah, S. Ahmad, and D. H. Kim, “Object detection mechanism based on deep learning
algorithm using embedded IoT devices for smart home appliances control in CoT,” J. Ambient Intell.
Humaniz. Comput., vol. 0, no. 0, p. 0, 2019, doi: 10.1007/s12652-019-01272-8.
12. M. A. Khan et al., “Smart Android Based Home Automation System Using Internet of Things (IoT),”
Sustain., vol. 14, no. 17, pp. 1–17, 2022, doi: 10.3390/su141710717.
13. S. Garg, A. Yadav, S. Jamloki, A. Sadana, and K. Tharani, “IoT based home automation,” J. Inf. Optim.
Sci., vol. 41, no. 1, pp. 261–271, 2020, doi: 10.1080/02522667.2020.1721581.
14. D. Geneiatakis, I. Kounelis, R. Neisse, I. Nai-Fovino, G. Steri, and G. Baldini, “Security and privacy
issues for an IoT based smart home,” 2017 40th Int. Conv. Inf. Commun. Technol. Electron.
Microelectron. MIPRO 2017 - Proc., no. 1, pp. 1292–1297, 2017, doi: 10.23919/MIPRO.2017.7973622.
15. M. Alaa, A. A. Zaidan, B. B. Zaidan, M. Talal, and M. L. M. Kiah, “A review of smart home applications
based on Internet of Things,” J. Netw. Comput. Appl., vol. 97, pp. 48–65, 2017, doi:
10.1016/[Link].2017.08.017.
16. B. Dokhnyak and V. Vysotska, “Intelligent smart home system using amazon alexa tools,” CEUR
Workshop Proc., vol. 2917, pp. 441–464, 2021.
17. D. Marikyan, S. Papagiannidis, and E. Alamanos, “A systematic review of the smart home literature: A
user perspective,” Technol. Forecast. Soc. Change, vol. 138, no. June 2018, pp. 139–154, 2019, doi:
10.1016/[Link].2018.08.015.
18. A. MIHALACHE, “Wireless Home Automation System using IoT,” Inform. Econ., vol. 21, no. 2/2017,
pp. 17–32, 2017, doi: 10.12948/issn14531305/21.2.2017.02.
19. N. Tetteh and O. Amponsah, “Sustainable adoption of smart homes from the Sub-Saharan African
perspective,” Sustain. Cities Soc., vol. 63, no. August, p. 102434, 2020, doi:
10.1016/[Link].2020.102434.
20. S. K. Sharma, U. K. Lilhore, S. Simaiya, and N. K. Trivedi, “An improved random forest algorithm for
predicting the COVID-19 pandemic patient health,” Ann. Rom. Soc. Cell Biol., vol. 25, no. 1, 2021.
21. L. Hellerstein, “Learning with Maximum-Entropy Distributions ∗,” pp. 123–145, 2001.
22. J. Ali, R. Khan, N. Ahmad, and I. Maqsood, “Random forests and decision trees,” IJCSI Int. J. Comput.
Sci. Issues, vol. 9, no. 5, pp. 272–278, 2012.
23. M. A. Muslim, Y. Dasril, M. Sam’an, and Y. N. Ifriza, “An improved light gradient boosting machine
algorithm based on swarm algorithms for predicting loan default of peer-to-peer lending,” Indones. J.
Electr. Eng. Comput. Sci., vol. 28, no. 2, pp. 1002–1011, 2022, doi: 10.11591/ijeecs.v28.i2.pp1002-
1011.
24. D. A. McCarty, H. W. Kim, and H. K. Lee, “Evaluation of light gradient boosted machine learning
technique in large scale land use and land cover classification,” Environ. - MDPI, vol. 7, no. 10, pp. 1–
22, 2020, doi: 10.3390/environments7100084.
25. R. Gayathri, S. U. Rani, L. Čepová, M. Rajesh, and K. Kalita, “A Comparative Analysis of Machine
Learning Models in Prediction of Mortar Compressive Strength,” Processes, vol. 10, no. 7, 2022, doi:
10.3390/pr10071387.
26. Z. Feng, C. Xu, and D. Tao, “Historical Gradient Boosting Machine,” Epic Ser. Comput., vol. 55, pp.
68–80, 2018.
27. L. W. Santoso, R. Lim, and and K. Trisnajaya, “利用許可書_見学G_2.pdf,” J. Inf. Commun. Converg.
Eng., vol. 16, no. 1, pp. 60–65, 2018.
April 2023 | 154
