A SURVEY ON MACHINE LEARNING AND DEEP

LEARNING TECHNIQUES WITH INTRUSION DETECTION

SYSTEM

R. Rajeshwari1, R. Malavika2
1
Assistant professor Computer Application Department
Bishop Heber College, Tiruchirappalli
2
Student Computer Application Department
Bishop Heber College Tiruchirappalli
1
rajeshwari.ca@bhc.edu.in
2
malavikar29@gmail.com

ABSTRACT: Networks plays a critical role in current existence; network security has become
a crucial research place. An intrusion detection device (IDS)which is crucial cyber security
technique, monitors the state of software and hardware running in the network. Regardless of a
long-time development, existing IDSs still face challenges in enhancing the detection accuracy,
lowering the false alarm rate and detecting unknown assaults. To clear up the above troubles,
many researchers have focused on growing IDSs that capitalize on machine learning and deep
learning techniques. Machine learning methods can robotically find out the important variations
between normal data and abnormal data with high accuracy. In addition, machine learning has
robust generalizability, so they’re also in a position to discover unknown attacks. Deep leaning
is a branch of machine learning, whose overall performance is remarkable and has emerged as a
research hotspot. In this paper, the concepts of machine and deep learning with their applications
are introduced. We review the difference between machine and deep learning and why when to
choose which technique. We also discuss about machine leaning and deep learning capitalized
with IDS.

KEYWORDS: Machine learning (ML), deep learning (DL), Artificial intelligence (AI),
Intrusion detection system (IDS), network security.

1. INTRODUCTION
In recent decades, Artificial intelligence (AI) enhanced as a wanted subject both within and
outside of the scientific community. The coalesced impact of new computing resources and
techniques with the avalanche of large data sets, is transforming many research areas and may
lead to technological breakthroughs that can be used by billions of people. In the recent years,
machine learning and particularly its sub-field deep learning has seen impressive advances. An
plenty of technology and non technology-based papers have covered the topics of machine
learning (ML), deep learning (DL)[1]. Yet there still persists a uncertainty around Machine
Learning and Deep Learning. The phrases are extremely associated, but are not interchangeable.
Machine learning is a branch of study which allows machines to learn from data or experience
and make a prediction based on the experience. Machine learning is used anywhere from
automating mundane tasks to offering intelligent intuition, industries in each sector tries to asset
from it. We may already be employing a device that utilizes it.
Machine learning algorithms allows the computers to look out from data, even ameliorate
themselves, without being explicitly programmed. Machine learning (ML) is a category of
algorithms that permits software applications to become more accurate in predicting outcomes
without being explicitly programmed [2]. The primary premise of machine learning is to erect
algorithms that can collect input data and use statistical analysis to predict an output while
updating outputs as new data becomes available.
As shown in the Figure.1, Deep learning is a limb of machine learning. Deep learning is
additionally a sort of mimic of human brain. In deep learning, we don’t get to explicitly program
everything. The concept of deep learning is not new. It has been around for a few of years now.
It is on hype nowadays because earlier we didn’t have that much processing power and a ton of
data.
As the processing power increments rapidly, deep learning and machine learning came within
the picture. These terms often materialize to be they’re interchangeable buzzwords, consequently
its momentous to understand the differences. The simplest sense for understanding the difference
between machine learning and deep learning is to understand that “deep learning is machine
learning”.
More accurately, deep learning is considered an evolution of machine learning. It uses a
programmable neural network that permits machine to form précised resolutions without the
assistance from humans.
The motive of this paper is to launch the core idea and concepts of machine learning (ML) and
deep learning (DL) with their applications, and finally we compare the characteristics between
Machine Learning and Deep Learning, why we move from machine learning (ML) to deep
learning (DL), when and where we need to choose machine learning and deep learning.

Figure.1 Venn diagram represents the border family

2. LITERATURE SURVEY
[14] This paper presents the survey of numerous research efforts spared towards the development
of intrusion detection system based on machine learning techniques. The author also presented
the survey works in tabular forms for easy understanding. Technique hired, dataset used and the
parameters evaluated are mentioned. Current achievements and limitations in developing
intrusion detection system by machine learning and future directions for research are also given
on this paper
[21] This paper gives the survey on Machine learning (ML) and Deep learning (DL) methods for
network analysis of intrusion detection and gives a brief tutorial description of each ML/DL
strategies. The author also gives each method and summarizes them based on their temporal or
thermal correlations. Due to the fact that data are so important in ML/DL methods, so they also
describe some commonly used network datasets used in ML/DL, discusses the challenges of
using ML/DL for Cybersecurity and also provide suggestions for research directions.
[3] This article will explore the use of machine learning and its methodologies. It also describes
the fields of deep learning which is being exploited in many leading IT providers will also be
clarified and discussed in this paper.
[7] This paper provides a overview of Machine learning and Deep learning with their applications
and concepts. It also presents a review in past and future application domains, sub-domains and
applications of machine learning and deep learning.
[8] This paper proposes a taxonomy of IDS that takes data objects as the main dimension to
classify and summarize machine learning based and deep learning-based IDS literature. The
author believes that this type of taxonomy framework is fit for cyber security researchers. This
survey first clarifies the concept and taxonomy of IDSs. This paper also introduces the machine
learning algorithms frequently used in IDSs, metrics, and benchmark datasets. Combined with
the representative literature, they also provide the taxonomic system as a baseline and explain
how to solve key IDS issues with machine learning and deep learning techniques. Challenges
and future developments are also discussed by reviewing recent representative studies.
[13] This paper provides the significant literature survey on machine learning (ML) and Deep
learning (DL) techniques for network analysis of intrusion detection. In addition, presents a short
tutorial explanation on every ML/DL method. As data holds a significant position in ML/DL
methods, subsequently this paper highlights the datasets used in machine learning techniques
which are the primary tools for analyzing network traffic and detecting abnormalities. In addition,
they elaborate on the issues confronted in the use of ML/DL for cybersecurity and offer
recommendations for future studies.

3. MACHINE LEARNING
Machine learning address the query of how to erect computers that ameliorate spontaneously
through experience. Machine learning is one of the spheres in the current computing technology.
A plenty of explorations has been under taken to make machine intelligent. Learning is human’s
natural phenomenon which has been made a vital feature of the machine as well. It’s one of the
leading rapidly elevating technological field of study, lying at the bisection of computing and
statistics, and at the center of artificial intelligence and data science [3]. The modern progress in
machine learning has been piloting both by the developments of new learning algorithms and
theory and by the endless detonation in the availability of online data and low-cost computation.
Machine learning methods can spontaneously recognize the crucial ddistinct between usual data
and unusual data with peak accuracy. Machine learning is a kind of artificial intelligence
technique that can spontaneity finds useful information from enormous data sets. Machine
learning is an integral a part of artificial intelligence, which is employed to style algorithms
supported the data trends and historical relationships between data.
Machine learning based IDS can aattain adequate detection levels when sufficient training data
is out there. Machine learning hinged IDS don’t pivot laboriously on domain knowledge,
consequently they’re facile to style and construct.
Machine learning algorithms are often broken down into two broad categories they are
supervised and unsupervised models. Supervised learning methods are wont to find a selected
target, which must also exist within the data. The most categories of supervised learning are
classification and regression. Classification models often have a binary intend at times expressed
as a “yes” or “no”, A variation on this model is likelihood assessment during which the target is
how probably a replacement observation is to fall into a selected category. Regression models
always have a numeric target, they model the connection between a variable and one or more
independent variables. Unsupervised learning methods are used when there’s no specific target
to seek out. Their purpose is to form grouping within the data set or make observations about
similarities. Further interpretation would be needed to form any decisions on these results.
Unsupervised learning model has two categories they are clustering and dimension reduction.
Clustering models look for subgroups within a data set that share similarities. These natural
groupings are almost like one another, but different than other groups. They may or might not
have any actual significance. Dimension reduction models reduce the number of variables in a
data set by grouping similar or correlated attributes. Its important to note that individual models
are not necessarily used in isolation. It often takes a mixture of supervised and unsupervised
methods to unravel a data science problem.
The most common machine learning algorithms are Artificial Neural Networks (ANN), Support
Vector Machine (SVM), K-Nearest Neighbour (KNN), naive Bayes, Logistic Regression (LR)
and K-Means. Traditional machine learning models are also known as shallow models [4].
The chore of machine learning is to distil worthy information from data, consequently the
operation of machine learning hinge upon the standard of the input data. Comprehension data
is the base of machine learning methodology. Traditional machine learning models cannot
address face to face data flow, therefore feature engineering is an vital notch before these
models can be applied. Feature engineering-based methods espouse a “feature vectors +
shallow models” mode. The feature vectors are suitable for many machine learning algorithms.
Combining domain knowledge with machine learning can improve the detection effect,
especially when the goal is to acknowledge certain types of attacks in specific application
scenarios.
Machine learning proffer the reassurance of having the ability to get précised models swiftly
and automatically. To date, materials scientists have used machine learning to build predictive
models for a couple of applications.

3.1 MACHINE LEARNING APPLICATIONS:

One of the well-liked applications of AI is machine learning (ML), in which computers,
software, and devices executes via cognition (very alike to human brain). Machine learning is
actively being utilized today, possibly in numerous places than one would expect. We probably
make use a learning algorithm dozen of time without even knowing it. Nowadays companies
make use of machine learning to upgrade business decisions, expand productivity, identify
disease, forecast weather and do many more things [5]. With the exponential development of
technology, we not only require superior tools to understand the data we currently have, but
we also require to prepare ourselves for the data we will have. To attain this goal, we require
to assemble intelligent machines. We can write a program to do basic things. But for most of
times hardwiring intelligence is hard. Finest way to do it is to have some way for machines to
learn and understand things themselves. A mechanism for learning - if a machine can learn
from input it does the work for us. This is where machine learning comes into action. Some
examples of machine learning are database mining for the development of automation,
applications that cannot be programmed. Traffic alerts, web search engine, photo tagging
applications, spam detector, social media, transportation and commuting, product
recommendations, virtual personal assistants, self-driving cars, dynamic pricing, Google
translate, online video streaming, fraud detection are some of the fields where machine learning
applications are used.

4. DEEP LEARNING
Deep learning is a derivative of machine learning, whose accomplishments astounding and has
become a research hot spot. The deep learning conceptualization materialized for the first time
in 2006 as a new field of research within machine learning. Deep learning is certainly a fast-
growing application of machine learning [6]. Deep learning is also an artificial intelligence
function that mock the workings of the human brain in processing data for use.
Deep learning AI are capable to acquire knowledge without human supervision, drawing from
data that is both unstructured and unlabeled. Deep learning takes part in an increasingly
dominant mantle in our lives. As the name deep learning connotes, it comprised of higher or
deeper number of processing layers, which disparity with shallow learning model with fewer
layers of units.
Deep learning methods have extremely impressive performance. Compared with traditional
machine learning techniques, deep learning methods are finer will handling with big data. Deep
learning methods can spontaneously learn feature representations from raw data. Adapt from
shallow to deep learning has authorized for further complexity and non-linear functions to be
mapped, as they can’t be efficiently mapped with shallow architectures. The main value of
deep learning relies on the optimization of existing application in machine learning.
Deep learning models have strong fixing potential and these models are black boxes. They
work in an end-to-end manner and are empirical. Deep learning is the deep structure which
contains multiple layers. Deep learning-based IDS has proliferated rapidly from 2015 to the
present. Deep learning can be attributed to the explosion of data or big data in the last 10 years.
Deep learning is feasibly the lion’s share within the development of computing field within the
recent times. Its influence has been sensed in nearly all scientific fields. There’s a contest
among the world’s leading economies and technology companies to progress deep learning.
Previously there are many areas where deep learning has exceeded human level capability and
performance.
Deep learning models can directly work on data [11]. Deep learning models directly learn
feature representations from the primary data. For larger data sets, deep learning methods have
a significant advantage over shallow learning.
Deep learning algorithms are a branch of the machine learning algorithms,
Deep learning models can be categorized into two they are supervised learning model and
unsupervised learning model. Deep Brief Networks (DBN), Deep Neural Networks (DNN ’s),
Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN ’s) comes under
supervised learning model. Auto-encoders, Restricted Boltzmann Machines (RBM ’s) and
generative adversarial networks (GAN) comes under the unsupervised learning models in deep
learning.

4.1 DEEP LEARNING APPLICATIONS:

A few years ago, we would’ve not even assumed that deep learning applications will bring us
self-driving cars and virtual assistants like Alexa, Siri and Google Assistant. But nowadays
these creations are a part of our day-to-day life [9]. Deep learning maintains to enchant us with
its countless possibilities like fraud detection and pixel restoration.
Deep learning applications may look enlighten to normal persons but those with the advantage
of knowing the machine learning world understand the reduction that deep learning is building
globally by exploring and resolving human difficulties in each domain.
In today’s world deep learning applications are used across varies industries. Self-driving cars,
new aggregation and fraud news detection, natural language processing, virtual Assistants,
entertainment, visual recognition, fraud detection, health care, personalization, detecting
development delay in children, colorizations of images, adding sounds to silent movies,
automatic machine translation, automatic handwriting generation, automatic game playing,
language translations, pixel restoration, photo descriptions, demographic and election
predictions, deep dreaming are a variety of the applications of deep learning[12].

5. MACHINE LEARNING AND DEEP LEARNING IN IDS :

An IDS is a sort of detection system that performs a key role in shielding cyber security by
monitoring the state of software and hardware running in a network. The first intrusion
detection device was proposed in 1980. since then, many mature IDS products have risen. But,
many IDSs still suffer from a excessive false alarm rate, producing many signals for low
nonthreatening conditions, which raises the load for security analysts and might cause severely
harmful attack to be unnoticed. Consequently, many researchers have focused on growing
IDSs with higher detection rates and reduced false alarm rates. Another problem with existing
IDSs is that they lack the capacity to detect unknown attacks. Because network environments
change quickly, attack variants and novel attacks emerge continuously. As a result, it is
essential to develop IDSs which can discover unknown unknown attacks. To cope with the
above issues, researchers have started to focus on building IDSs using machine learning
methods. Machine learning is a type of artificial intelligence approach that could robotically
discover beneficial information from big datasets [8]. Machine learning-based IDSs can obtain
satisfactory detection levels when sufficient training data is available, and machine learning
models have sufficient generalizability to stumble on attack variants and novel attacks. Further,
machine learning based IDSs do not depend heavily on domain knowledge. Therefore, they
are easy to design and construct [10]. Deep learning is a limb of machine learning that may
reap awesome performances. In comparison with traditional machine learning techniques, deep
learning is better at dealing with large datasets. Moreover, deep learning strategies can
mechanically learn feature representations from raw data [15]. They function in an end-to-end
style and are practical [14].

5.1 MACHINE LEARNING BASED IDS:

Table.1 represents the survey on papers based on machine learing based IDS
Paper Author Machine learning Datasets Methods
Algorithms

[22] Zouhair Chiba, DNN CICIDS 2017, Network intrusion

Noreddine Abghour, GA NSL-KDD version detection system
Khalid Moussaid, SAA 2015 and CIDDS-
Amina EI omri, 001
Mohamed Rida.

[17] Indrajeet Kumar SVM KDD-99 Principal

KNN component analysis

[16] Prachi shukla K-Means Private dataset Wormhole attack

Decision tree detection

[23] Abhishek Verma , CIDDS-001, Performance

Virender Ranga UNSW-NB15, analysis
NSL-KDD

[24] Anish Halimaa A, SVM, Naive NSL-KDD Comparative

Sundarakantham .K Bayes analysis

5.2 DEEP LEARNING BASED IDS:

Table.2 represents the survey on papers based on deep learning-based IDS
Paper Author Deep learning Datasets Methods
Algorithms
[25] Zheng Wang State-of-the-art attack NSL-KDD Perturbatio
n methods
[26] Chuanlong RNN, CNN NSL-KDD Accuracy
Yin, Yuefie analysis
Zhu, Xinzheng
He
[27] Zhiqiang Liu, DNN NSL-KDD Gateway
Mohi- Ud- analysis
Din-
Ghulam
[19] Aechan Kim, CNN Public Datasets Accuracy
Mohyun Park, CSIC-2010, CICIDS2017 analysis
Dong Hoon
Lee
[28] Yazan Otoum, SMO, SDPN DoS, U2R, R2L Extensive
Dandan Liu, analysis
Amiya Nayak

6. COMPARING MACHINE LEARNING WITH DEEP LEARNING:

Shallow models compared to deep models, deep learning is a limb of machine learning, and
the consequence of deep learning models are evidently higher-ranking to those of the
traditional machine learning methods in most application frameworks.
Compared with shallow machine learning models, deep learning models posses stronger fitting
and generalizeed potential. Additionally, deep learning strategies are independent of feature
engineering and domain knowledge, which takes a smashingg profitability over shallow
learning models [7]. However, the running time of deep learning models are often too long to
meet the real-time requirements. The chief dissimilarity between deep learning and traditional
machine learning is its performance as the quantity of data enlarges.
Deep learning algorithms do not work well when the data volumes are small, because deep
learning algorithms demands a huge number of data to recognize the data flawlessly.
Conversely, in this case, when the traditional machine-learning algorithm uses the established
rules, the potential will be superior.
The Deep learning algorithm demands many matrix operations. The GPU is predominantly
used to optimize matrix operations efficiently. Therefore, the GPU is the hardware mandatory
for the Deep learning to work accurately. Deep learning depends more on high-performance
machines with GPU ’s than do traditional-learning algorithms.
As shown in Figure.2, feature processing is the operation of putting domain knowledge into a
feature extractor to lessen the difficulty of the data and generate patterns that make learning
algorithms work superior. Feature processing is time consuming and requires specialized
knowledge. In machine learning, most of the characteristics of an application must be
determined by an expert and then encoded as a data type. The accomplishment of most
Machine learning algorithms pivot upon the precision of the features extracted. Attempting to
acquire high level features directly from data is a prime difference between Deep learning and
traditional machine learning algorithms. Thus, Deep learning tapers the struggle of designing
a feature extractor for each problem.
Will attempting traditional machine learning algorithms to figure out the problems, traditional
machine learning normally shatters the problem into numerous sub-problems and resolves the
sub-problems, eventually acquiring the conclusive consequence. On the contrary, deep
learning advocates direct end to end problem solving.
The time period includes both training and test time. Due to the acme of convolution in deep
models, both their training and test times are lengthy than those of shallow models. Generally,
it takes a long time to train a Deep learning algorithm because there are many parameters in
Deep Learning algorithm, consequently, the training steps are prolonged. The most important
advanced Deep learning algorithm, like ResNet, takes accurately two weeks to complete a
training session, whereas Machine learning training takes relatively little time, only seconds to
hours. Though, the test time is literally the reverse. Deep learning algorithms requires
extremely small time to run during testing contrasted with some Machine learning algorithms,
the test time enlarges as the amount of data enlarges. Even so this point does not involve to all
Machine learning algorithms, as some Machine learning algorithms do have short test
times[13].
Vitally, interpretability is an chief element in differentiating Machine learning with deep
learning. Machine learning algorithm supplies crystal clear rules for why the algorithm chooses
so, consequently it is effortless to describe the reasoning behind the decision. Deep learning
models are black boxes. The result are almost uninterpretable, which is a censorious point in
deep learning.

Figure.2 represents the process in machine learning and deep learning while reading a data

7. WHY FROM MACHINE LEARNING TO DEEP LEARNING

Machine learning and deep learning have embellished has two major sought after developing
technologies. It is no further a hint anymore. Its been turn out into an actuality. Tech monsters
like Amazon, Facebook, Google are utilizing machine and deep learning to relent human tasks
and automate anything and everything. Deep learning is obtaining a lot of frame due to its
superiority in terms of exactness when trained with enormous amounts of data.
Presently software industries proceed towards machine intelligence. Machine learning has
become essential in each sector as a way of making machines smarter. In a simpler way,
machine learning is a cluster of algorithms that parse data, learn from them, then apply what
they’ve learned to form intelligent decisions. Example of machine learning are everywhere. Its
how Netflix knows which show we want to view next or how Facebook recognizes the friend’s
face in a digital photo. The thing about traditional machine learning algorithms is that as
complex as they’ll seem, they’re still machine like. They need lot of domain proficiency;
human mediation is only capable of what they’re designed for nothing more or nothing less
[16]. For AI developer and the rest of the world, that’s where deep learning holds a bit more
promise.
Practically, deep learning is a segment of machine learning that achieves enormous volume
and versatility by learning to illustrate the world as nested hierarchy of concepts, with each
concept explained in regards to fundamental ideas, and more abstract representations computed
in terms of less abstract ones.
Deep learning demands high-end machine contrary to traditional machine learning algorithms.
GPU has become a essential part now to perform any deep learning algorithm [17].
In traditional machine learning techniques, nearly all of the applied features need to be
identified by a domain expert in order to lower the difficulty of the data and frame patterns
further visible to learning algorithms to work. The greatest benefit of deep learning algorithms
is as they try to learn high level features from data in an incremental manner[18]. This
eliminates the necessity of domain expertise and hardcore feature extraction.

8. CHOOSING BETWEEN MACHINE LEARNING AND DEEP LEARNING:

Machine learning provides a variety of techniques and models we can choose based on the
application, the sized of data we’re processing, and the type of problem we want to solve [19].
A successful deep learning application demands a great amount of data (thousands of images)
to train the model, as well as GPU or graphics processing units, to quickly process the data.
When choosing between machine learning and machine learning, examine whether we have
high performance GPU and lots of labeled data. If we don’t have either of those things, it may
make more sense to use machine learning instead of deep learning [20]. Deep learning is
normally more complex, so we’ll need at least a few thousand images to get definitive results.
Having a excessive performance GPU means the model will take less time to examine all those
images.

9. CONCLUSION:
This paper has specified all the fundamentals regarding machine learning and deep learning,
and a glance at machine learning and deep learning applications. We concentrated on machine
learning and deep learning to find out the difference. Their functioning is utterly similar, but
they are distinct in final capabilities. Machine learning can make the conclusion from the
learned data but with the certain assistance of human beings. Machine learning predicts the
results, the executive core of the human specialists will be to illuminate and help them to
execute with greater accuracy. Anyway, it is the optimization of machine learning performance.
Concerning deep learning, the process of making conclusions goes much further. Its algorithms
behave similarly to the human brain. The structure of deep learning furnish the possibility to
analyses a big amount of the data while being educated by it for the performance of the various
tasks.
Both machine learning and deep learning examines the data and learn from it, but only deep
learning attempts to replicate the activities of the human brain when it has to make the
conclusion.
The continuous evolution of AI permits new possibilities for machine development. Deep
learning and machine learning are considered to be the subcategories of Artificial intelligence.
Both machine learning and deep learning are the special algorithms which will perform certain
tasks, distinguished by their own advantages. We have understood what exactly are machine
learning and deep learning, and found the difference between them.

REFERENCES:
1. Xin, Y., Kong, L., Liu, Z., Chen, Y., Li, Y., Zhu, H., ... & Wang, C. (2018). Machine learning and deep
learning methods for cybersecurity. Ieee access, 6, 35365-35381.
2. Zhang, Lu, et al. "From machine learning to deep learning: progress in machine intelligence for rational
drug discovery." Drug discovery today 22.11 (2017): 1680-1685.
3. Ongsulee, Pariwat. "Artificial intelligence, machine learning and deep learning." 2017 15th International
Conference on ICT and Knowledge Engineering (ICT&KE). IEEE, 2017.
4. Soofi, Aized Amin, and Arshad Awan. "Classification techniques in machine learning: applications and
issues." Journal of Basic and Applied Sciences 13 (2017): 459-465.
5. Kolouri, Soheil, et al. "Optimal mass transport: Signal processing and machine-learning
applications." IEEE signal processing magazine 34.4 (2017): 43-59.
6. Goodfellow, Ian, et al. Deep learning. Vol. 1. No. 2. Cambridge: MIT press, 2016.
7. Shinde, Pramila P., and Seema Shah. "A review of machine learning and deep learning
applications." 2018 Fourth International Conference on Computing Communication Control and Automation
(ICCUBEA). IEEE, 2018.
8. Liu, Hongyu, and Bo Lang. "Machine learning and deep learning methods for intrusion detection systems:
A survey." applied sciences 9.20 (2019): 4396.
9. Deng, Li, and Dong Yu. "Deep learning: methods and applications." Foundations and trends in signal
processing 7.3–4 (2014): 197-387.
10. Pouyanfar, Samira, et al. "A survey on deep learning: Algorithms, techniques, and applications." ACM
Computing Surveys (CSUR) 51.5 (2018): 1-36.
11. Goodfellow, Ian, Yoshua Bengio, Aaron Courville, and Yoshua Bengio. Deep learning. Vol. 1, no. 2.
Cambridge: MIT press, 2016.
12. Dargan, Shaveta, et al. "A survey of deep learning and its applications: A new paradigm to machine
learning." Archives of Computational Methods in Engineering (2019): 1-22.
13. Salloum, Said A., et al. "Machine learning and deep learning techniques for cybersecurity: a review." Joint
European-US Workshop on Applications of Invariance in Computer Vision. Springer, Cham, 2020.
14. Hamid, Yasir, M. Sugumaran, and V. R. Balasaraswathi. "Ids using machine learning-current state of art
and future directions." Current Journal of Applied Science and Technology (2016): 1-22.
15. Aminanto, Erza, and Kwangjo Kim. "Deep learning in intrusion detection system: An overview." 2016
International Research Conference on Engineering and Technology (2016 IRCET). Higher Education Forum,
2016.
16. Shukla, Prachi. "ML-IDS: A machine learning approach to detect wormhole attacks in Internet of
Things." 2017 Intelligent Systems Conference (IntelliSys). IEEE, 2017.
17. Kumar, Indrajeet, et al. "Development of IDS Using Supervised Machine Learning." Soft Computing:
Theories and Applications. Springer, Singapore, 2020. 565-577.
18. Damopoulos, Dimitrios, et al. "Evaluation of anomaly‐based IDS for mobile devices using machine
learning classifiers." Security and Communication Networks 5.1 (2012): 3-14.
19. Kim, Aechan, Mohyun Park, and Dong Hoon Lee. "AI-IDS: Application of deep learning to real-time Web
intrusion detection." IEEE Access 8 (2020): 70245-70261.
20. Thakkar, Ankit, and Ritika Lohiya. "A Review on Machine Learning and Deep Learning Perspectives of
IDS for IoT: Recent Updates, Security Issues, and Challenges." Archives of Computational Methods in
Engineering (2020): 1-33.
21. Xin, Yang, et al. "Machine learning and deep learning methods for cybersecurity." Ieee access 6 (2018):
35365-35381
22. Chiba, Zouhair, et al. "Intelligent approach to build a Deep Neural Network based IDS for cloud
environment using combination of machine learning algorithms." Computers & Security 86 (2019): 291-317.
23. Verma, Abhishek, and Virender Ranga. "Machine learning based intrusion detection systems for IoT
applications." Wireless Personal Communications 111.4 (2020): 2287-2310.
24. Halimaa, Anish, and K. Sundarakantham. "Machine learning based intrusion detection system." 2019 3rd
International conference on trends in electronics and informatics (ICOEI). IEEE, 2019.
25. Wang, Zheng. "Deep learning-based intrusion detection with adversaries." IEEE Access 6 (2018):
38367-38384.
26. Yin, Chuanlong, et al. "A deep learning approach for intrusion detection using recurrent neural
networks." Ieee Access 5 (2017): 21954-21961.
27. Liu, Zhiqiang, et al. "Deep learning approach for IDS." Fourth International Congress on Information and
Communication Technology. Springer, Singapore, 2020.
28. Otoum, Yazan, Dandan Liu, and Amiya Nayak. "DL‐IDS: a deep learning–based intrusion detection
framework for securing IoT." Transactions on Emerging Telecommunications Technologies (2019): e3803.