MACHINE LEARNING APPLICATIONS IN MALWARE CLASSIFICATION: A METAANALYSIS LITERATURE REVIEW

International Journal on Cybernetics & Informatics (IJCI) Vol. 12, No.1, February 2023
David C. Wyld et al. (Eds): NLDML, CCITT, EDUTEC, AUEN -2023
pp. 113-124, 2023. IJCI – 2023 DOI:10.5121/ijci.2023.120109
MACHINE LEARNING APPLICATIONS IN
MALWARE CLASSIFICATION: A META-
ANALYSIS
LITERATURE REVIEW
Tjada Nelson, Austin O’Brien and Cherie Noteboom
Beacom College of Computer & Cyber Sciences, Dakota State University,
Madison, South Dakota
ABSTRACT
With a text mining and bibliometrics approach, this study reviews the literature on the evolution
of malware classification using machine learning. This work takes literature from 2008 to 2022
on the subject of using machine learning for malware classification to understand the impact of
this technology on malware classification. Throughout this study, we seek to answer three main
research questions: RQ1: Is the application of machine learning for malware classification
growing? RQ2: What is the most common machine-learning application for malware
classification? RQ3: What are the outcomes of the most common machine learning
applications? The analysis of 2186 articles resulting from a data collection process from peer-
reviewed databases shows the trajectory of the application of this technology on malware
classification as well as trends in both the machine learning and malware classification fields of
study. This study performs quantitative and qualitative analysis using statistical and N-gram
analysis techniques and a formal literature review to answer the proposed research questions.
The research reveals methods such as support vector machines and random forests to be
standard machine learning methods for malware classification in efforts to detect maliciousness
or categorize malware by family. Machine learning is a highly researched technology with
many applications, from malware classification and beyond.
KEYWORDS
Malware, Malware Classification, Machine Learning.
1. INTRODUCTION
Machine learning is a technology that has been at the forefront of academic research since its
inception in the 1990s [1]. Applications from Machine learning is defined as the capacity for a
system to learn from a problem-specific data source to identify patterns in that data build that
provide insight around that data [2]. The technology behind machine learning enables a wide
range of efficiency for computer systems across many disciplines. Therefore, machine learning
has been the subject of many papers in academia. As of December 2022, Google Scholar, an
academic article search engine and aggregator, returns over 5.4 million results for the term
“machine learning.” Through this research, machine learning has successfully been applied to
several real-world applications ranging from healthcare to gaming.
Machine learning can be divided into three types: supervised, unsupervised, and reinforcement.
Supervised machine learning takes test data representing known desired results and is used to

114
produce a system that predicts future results. For example, supervised machine learning is the
method to perform classification or face recognition. Unsupervised machine learning takes data
and tries to identify clusters to classify similar or related data points and insights. This machine
learning type builds recommendation systems, anomaly detection, and tracking buy habits in
customer transactions. Reinforcement machine learning allows the system to operate and then
notify it when it makes mistakes, so it learns to avoid them. This machine learning type is used to
create video game AI and operation simulations. Machine learning has proved to be very useful
in solving real-world problems in all these application types.
A significant problem in cybersecurity is malware which plays a major role in cybercrime. AV-
TEST institute reported more than 1 billion infected files in 2021, showing the sheer volume of
malware cybersecurity analysts contend with year over year [3]. Malware, by definition, cause
harm to systems and inflict other damages, such as financial costs, to victims. Cybercrime costed
victims around $6 Trillion USD in 2021, further highlighting the need for this problem to be
addressed [4]. The term malware is derived from the words malicious, and software based on its
usage to cause damage on a target system. Malware is used by cybercriminals to conduct
operations and run illicit businesses focused on the perpetration of cybercrime [6].
Analyzing malware allows defenders to detect infections and detect protections. There are two
types of analysis commonly used to investigate malware, static and dynamic. Static analysis
entails investigating a malware sample without executing it while dynamic analysis is observing
the malware activity while it is running [7]. There are various techniques used for static analysis
such as file fingerprinting, extraction of hard coded strings, file format, anti-virus scanning,
packer detection and disassembly [8]. These static analysis techniques rely on the binary
representation of the program to provide a safe and manual approach to analysis. File
fingerprinting and file format techniques provide characteristics about the program. anti-virus
scanning and packer detection offer identification and classification of the program and its
contents. Hard coded strings and disassembly offer detailed extraction of data harboured within
the program. Static analysis could provide insight on the information that is missed during
dynamic analysis do to program paths not executed [9]. It allows for the classification of these
malware to get a general understanding of the nature of the malware versus an in-depth analysis
of the binary.
Ultimately allowing defenders to determine if a binary is malicious or not and then use that
information to protect their systems. Although static analysis is generally safer than dynamic
analysis the original source code of the analyzed program is usually not available. Techniques
such as machine learning against this static information provides additional efficiencies in the
analysis of these binaries. The malware must be detected to prevent or recover from malware
infections. One method of this detection is malware classification, which identifies malicious
software and its functionality from the assortment of software present on a system.
Understanding if malware is on a system and the functionality of that malware helps an analyst
provide the proper corrective actions to minimize the risk of further damage incurred by the
malicious event. Machine learning is particularly suited for this type of classification and has
been applied in mass to the problem of malware classification.
However, with the wide variety of machine learning applications, the impact of machine learning
on this problem is still in question. Wagstaff proposes three focus areas to address the impact of
machine learning which include benchmarking of datasets, performance measurements, and how
it is applied [5]. These focus areas raise legitimate questions about how machine learning is used
for malware classification and the impact of these applications. Thus the motivation for this study
is to survey the current literature in the field of malware classification and machine learning to

115
identify the most common uses and applications. This would contribute to the current body of
knowledge by building a foundation for future research around the current applications
To this end, this study will survey the volume of machine learning literature directed toward
malware classification. Furthermore, this study aims to gather data to answer the following
research questions: 1. Is the application of machine learning for malware classification growing?
2. What is the most common machine learning application used for malware classification? 3.
What are the outcomes of the most common machine learning applications? Using the
information extracted from this research to answer the above questions will also provide an up-
to-date state-of-the-art on machine learning in malware classification applications.
2. METHODOLOGY
2.1. Search Strategy and Data Collection
We have defined three research questions to guide our meta-analysis of research publications on
machine learning and malware classification. We then developed search queries targeting four
databases, Association for Computing Machinery (ACM), Institute of Electrical and Electronics
Engineers Xplore (IEEE Xplore), Science Direct, and Web of Science. The query leveraged for
these databases includes any references to the terms “machine learning” and “malware
classification” in the metadata of a publication (“All Metadata”: malware classification) AND
(“All Metadata”: machine learning). The results from this query were filtered down to
publications with dates from 2017 to September 2022. The reason for this constraint allows the
research to focus on the current state versus historical applications.
The authors reviewed the titles of the articles from this 2186 sized article dataset to determine the
relevance of the titles toward malware classification and machine learning, removing any
outliers. 372 articles were removed, leaving a total of 2186 unique articles gathered. Based on
additional searches performed authors were able to determine these search times and filters
provided the most suitable dataset for performing meta-analysis against terms in the article
metadata. Metadata from these articles were collected, including but not limited to publication
title, authors, abstract, keywords, and publication year. These fields were extracted by importing
all the dataset articles into the citation tool Zotero by exporting all of the articles to a CSV
format. The collected metadata enables this research to focus on the articles most relevant to
answering the proposed research questions. Beyond this study, this article collection will be used
in future literature reviews.
2.2. Data Analysis
From this metadata collection, we used the text mining process described in figure 1 to complete
the analysis. This text mining process was informed by the work of Zheng et. al. however the
implementation of this mining process was executed by python instead of Vos viewer software
[10]. All stop words were removed, and the metadata was tokenized to prepare word cloud and
N-gram analysis. With more insight into the trends and themes of the dataset, the N-gram
analysis was leveraged to identify terms of interest, such as machine learning applications for
literature review.

116
Figure 1. Text Mining Process.
3. RESULTS
3.1. Exploratory Data Analysis
Initial descriptive data analysis was conducted using our article data set to gain insights around
current trends. This analysis included 1. Determining the number of articles published per year, 2.
Calculating word frequency of top ten words, 3. Generating a word cloud, and 4. Performing N-
gram analysis.
Figure 2 shows the number of articles published each year from 2017 to September 2022. Based
on these yearly publishing numbers, there is a clear trend upward year over year, with a drop in
2022 due to the year not being completed by the time of data collection. The upward trendline
answers our first research question: R1. Is the application of machine learning for malware

117
classification growing? Based on these results, we see a clear upward trend for the application of
machine learning for malware classification.
Figure 2. Number of articles published in each year.
A subsequent review of the top 10 words in the title field of metadata for the collection after
removing stop words using the python libraries Word Cloud and NLTK used for stop words was
performed. Table 1 shows the top 10 words from this data set. Looking at the results, ‘malware,’
‘learning,’ and ‘classification’ were some of the most used terms. Terms such as ‘detection’ are
also listed in this top 10 list, which could suggest that a detection system is the most common
result of machine learning for malware classification.
Table 1. Top 10 most frequent title words.
Word Count
malware 812
learning 735
detection 673
using 541
classification 503
machine 406
android 302
based 297
network 280
deep 262
Another version of word frequency analysis conducted in this study is the Top 10 most frequent
words in the abstracts of the collected data. Table 2 shows the results of this analysis, were as
expected, ‘malware,’ ‘machine,’ ‘learning,’ and ‘classification’ are all frequently used. In
addition, as identified with the title frequency, ‘detection’ is also frequently used.

118
Table 2. Top 10 most frequent abstract words.
Word Count
malware 5075
feature 3473
learning 3413
model 2911
classification 2825
detection 2593
method 2494
machine 2253
data 2213
network 1920
Word clouds provide an opportunity to visually observe the more extensive frequency data set,
which can lead to additional findings that were not readily apparent in the previous tables. Word
clouds represent volume by the size of the specific term instead of a word count value. This
approach provides a visual cue for trends and insights. Figure 3 shows a word cloud of the most
frequently used words in the abstracts of the data set. We can confirm that ‘malware,’ ‘feature,’
‘learning,’ and ‘model’ are the most frequently used words based on the size of the associated
text of the word.
Figure 3. Word cloud of most frequent abstract words.
3.2. N-Gram Analysis of Abstract Terms
N-gram analysis takes combinations of terms within a text dataset to determine the frequency of
pairs of words seen together. The results of this type of analysis allow the researcher to observe
more sophisticated themes from a dataset. The aggregated abstract data from this study was
placed in the NLTK n-gram library to identify these patterns in bi-gram, trigram, and 4-gram
analysis. Table 3 displays the top 20 bi-gram terms within the abstract words. As expected, sets
such as (“machine,” “learning,”) or (“malware,” “detection”) have the highest count. Looking
down at this table, we find sets that provide insight around machine learning applications such as
(“neural,” “network”), (“deep,” “learning”), (random, forest), (support, vector), (vector, machine)

119
and (convolutional, neural). These terms suggest the types of machine learning applications used
within malware classification. This dataset also suggests that android malware is a prominent
subject of this type of study based on the counts within the n-gram analysis.
Table 3. Top 20 bi-gram analysis from abstract words.
Bi-gram Words Count
(machine, learning) 1695
(malware, detection) 969
(neural, network) 706
(deep, learning) 581
(android, malware) 437
(experimental, result) 388
(learning, algorithm) 383
(result, show) 368
(learning, model) 363
(malware, classification) 300
(random, forest) 296
(feature, selection) 263
(learning, technique) 260
(support, vector) 258
(proposed, method) 249
(vector, machine) 244
(malware, family) 239
(learning, method) 223
(convolutional, neural) 220
(classification, accuracy) 207
Table 4 provides a more straightforward depiction of these sets of terms by showing the top 10
tri-grams from the abstract words in the dataset. Finally, looking at the machine learning
applications inferred from this table data, (“support,” “vector,” “machine”) and (“convolutional,”
“neural,” “network”) are presented as prominent applications for malware classification using
machine learning.
Table 4. Top 10 tri-gram analysis from abstract words.
Tri-gram Words Count
(machine, learning, algorithm) 272
(support, vector, machine) 239
(convolutional, neural, network) 218
(machine, learning, technique) 195
(machine, learning, model) 168
(experimental, result, show) 163
(android, malware, detection) 155
(deep, learning, model) 129
(deep, neural, network) 127
(machine, learning, classifier) 118

120
The four-gram analysis of the abstract words are displayed in Table 5. This table shows Support
Vector Machines and Convolutional Neural Networks at the dataset’s top four-gram and most
prominent machine learning application. Towards the bottom of the top 10 list are decision tree
random forests and recurrent neural networks, also machine learning applications.
Table 5. Top 10 four-gram analysis from abstract words.
Four-gram Words Count
(support, vector, machine, svm) 88
(convolutional, neural, network, cnn) 81
(experimental, result, show, proposed) 35
(convolutional, neural, network, cnns) 28
(using, machine, learning, technique) 24
(machine, learning, deep, learning) 22
(decision, tree, random, forest) 21
(microsoft, malware, classification, challenge) 21
(support, vector, machine, classifier) 20
(recurrent, neural, network, rnn) 19
3.3. Distribution of Machine Learning Applications
Based on the N-gram analysis conducted during this study, we can address research question 2,
addressing the most common machine learning applications used for malware classification. The
study observes random forest decision trees (which will be shortened to random forest to
generalize this application), support vector machines, convolutional neural networks, and
recurrent neural networks as the most common applications of machine learning in malware
classification. Figure 4 shows the number of articles containing the designated terms, which
concluding support vector machines are the most common machine learning application for
malware classification.
Figure 4. Number of articles with the specific machine learning application

121
3.3.1. Random Forest
Random forest is a classifier based, supervised machine-learning algorithm which leverages
multiple decision trees to identify patterns [11]. Using this model takes a training set and a test
set to build a classifier for the model. In many of the applications observed in this study, static
features of the software are used to determine whether it is malware or not. In a study by Bowen
Sun et al., they used static features with several machine learning algorithms, including support
vector machines, decision trees, random forests and others, finding they were able to get the
highest F1 score which was the highest of all tests conducted [12]. The Zhang et al. study also
resulted in random forest as the highest-performing algorithm using the Ember dataset. The
Zhang et al. study also noted that random forest is the most cost-effective algorithm they tested
[13]. The dataset from this study found a few articles that targeted this approach against android
malware versus the traditional computer system malware; a finding supported by the N-gram
analysis findings. In total, 66 articles from the dataset specifically call out random forest in the
abstract, making it the second most common machine learning application for malware
classification.
3.3.2. Support Vector Machine
Support vector machines (SVM) is a learn-by-example machine learning algorithm, essentially
using a large dataset of a particular pattern to determine how to find that pattern. It is very good
at binary classification but struggles with granularity [14]. In our use case, SVM would be useful
for identifying if a binary was malicious but not the category of maliciousness of the binary.
Gravilut et. al, achieved a 99% accuracy and 0.75% false positive rate using a Boosted SVM
model using static features of the binary as their data set. Wang and Wu were able to generate
promising results by using SVM to detect packed executables, a common characteristic of
malware; this study offered a use case for malware classification that differed from static feature
extraction [15]. In the study “Malware Family Classification using Active Learning by Learning”
by Chen et al., the learning algorithms using byte and assembler data from binaries were used to
create an SVM model. The group found that using SVM with Active Learning by Learning
(ALBL) could effectively perform malware classification [16]. Furthermore, 103 articles from the
dataset directly mentioned SVM, the highest frequency count of the terms explored. Based on
this number, SVM would be the most common machine learning application for malware
classification, addressing research question 2.
3.3.3. Convolutional Neural Network
Based on early findings from the study of biological vision, convolutional neural networks
(CNN) uphold that notable aspect of the neural network’s data set, making it a good solution for
image and natural language processing [17]. In our dataset, only 9 articles directly mention CNN
as a model in their abstract. This image-based processing application represented several of the
articles in the data set. Fathurrahman et al., leveraged CNNs to perform malware classification on
embedded systems using images of malware from the Malevis dataset, finding an approach that
was 19.22% accurate and could be run on an embedded IoT system effectively [18]. Another
notable article from the dataset containing CNN in the abstract is “Malware Classification using
Deep Convolutional Neural Networks” by Kornish et. al, found using transfer learning to retain
the deep convolutional neural network. However, they noted that the article’s architecture had
little practical application [19]. Lin et. al, used RNN to classify IoT malware in their study; this
research used opcode sequence to detect patterns using RNN.

122
3.3.4. Recurrent Neural Network
Recurrent neural networks (RNN) have a simple neural network structure with a built-in feedback
loop allowing it to be used for forecasting. This algorithm is typically used to detect patterns in a
data sequence [20]. For example, with malware classification, Shukla et. al, took the approach of
visualizing the behavior of the malware in the form of a heat map and leveraging RNN to achieve
a 94% accuracy score. There were 37 articles in the dataset which mentioned RNN in the abstract
of the articles around malware classification, and images were used as the dataset in the machine
learning process.
3.4. Research Themes
During the research of this study, Android malware emerged as a trending topic in the dataset. A
significant number of articles in the data set showed up in the top 5 frequently used bi-grams
from the n-gram analysis. Mobile malware, including Android-based malware, has grown in
market share compared to other malware types despite the general malware decline [21]. The
shift in the malware landscape and the size of the mobile market are apparent trends. These trends
could explain why android malware classification solutions are turning to machine learning
increasingly.
Several articles extracted static features of malware using pe and malware analysis tools to create
a dataset for machine learning. Another feature set for datasets was images of the binary using
processing tools. These two feature sets were the most prevalent within the dataset versus others,
such as function call graphs and real-time activity. Future research in function call graphs and
real-time activity as machine learning features is worthwhile exploring in the future.
4. DISCUSSION
From the analysis of 2186 articles around the application of machine learning for malware
classification, it is evident that machine learning provides clear advantages in this task.
Furthermore, machine learning can be performed in a variety of ways using different algorithms
and datasets.
4.1. R1: Is the application of machine learning for malware classification growing?
This study's first research question sought to understand if the usage of machine learning for
malware classification is increasing. The study answers this question by creating the chart using
the distribution of articles by publication year within the dataset displayed in Figure 2. This chart
shows an apparent increase in publications year over year until 2022, which was the current year
at the time of the study. The information from this chart shows an apparent growth in articles
concerning malware classification and machine learning, answering research question 1.
4.1.1. R2: What is the most common machine learning application used for malware
classification?
This study used an N-gram analysis of abstract text terms against the dataset to understand the
more complex topics discussed within these articles. This research analysis revealed several
terms associated with machine learning applications and algorithms. Extracting the machine
learning algorithms from the N-gram results, the study determined the most common by string
searching those terms across the whole dataset. This effort found support vector machines (SVM)
as the most common application of machine learning in malware classification. 103 articles

123
within the dataset mentioned SVM in the abstract, with the following more frequent application
being random forest with 66 article mentions.
4.1.2. R3: What are the outcomes of the most common machine learning applications?
This study aimed to answer the third research question around identifying the outcomes of the
typical machine learning applications used for malware classification. This question would
provide insight into the contributions to the body of knowledge this area of study has provided.
Based on the literature reviews guided by the N-gram analysis results, these typical machine
learning applications' outcomes are malware detection and family categorization. The outcomes
were shared across all the machine-learning applications explored in this research. This outcome
was also prevalent in the title word analysis, with detection as one of the top terms present in a
title.
5. CONCLUSIONS
In conclusion, this study reviewed 2186 articles about malware classification and machine
learning. We found that machine learning for malware classification is steadily growing in
popularity. The most common machine learning applications used for this purpose are support
vector machines and other classifiers, primarily used for malware detection and malware family
categorization. These findings highlight the potential of machine learning as a powerful tool in
the fight against malware and highlight the importance of continued research in this area.
REFERENCES
[1] Ö. Çelik, “A Research on Machine Learning Methods and Its Applications,” J. Educ. Technol. Online
Learn., Sep. 2018, doi: 10.31681/jetol.457046.
[2] C. Janiesch, P. Zschech, and K. Heinrich, “Machine learning and deep learning,” Electron. Mark., vol.
31, no. 3, pp. 685–695, Sep. 2021, doi: 10.1007/s12525-021-00475-2.
[3] M. Asam et al., “Detection of Exceptional Malware Variants Using Deep Boosted Feature Spaces and
Machine Learning,” Appl. Sci., vol. 11, no. 21, p. 10464, Nov. 2021, doi: 10.3390/app112110464.
[4] F. A. Aboaoja, A. Zainal, F. A. Ghaleb, B. A. S. Al-rimy, T. A. E. Eisa, and A. A. H. Elnour,
“Malware Detection Issues, Challenges, and Future Directions: A Survey,” Appl. Sci., vol. 12, no. 17,
p. 8482, Aug. 2022, doi: 10.3390/app12178482.
[5] A. Submission, “Machine Learning that Matters,” p. 6.
[6] M. Lindorfer, A. Di Federico, F. Maggi, P. M. Comparetti, and S. Zanero, “Lines of malicious code:
insights into the malicious software industry,” in Proceedings of the 28th Annual Computer Security
Applications Conference on - ACSAC ’12, Orlando, Florida, 2012, p. 349. doi:
10.1145/2420950.2421001.
[7] A. Afianian, S. Niksefat, B. Sadeghiyan, and D. Baptiste, “Malware Dynamic Analysis Evasion
Techniques: A Survey.” arXiv, Nov. 03, 2018. Accessed: Jan. 11, 2023. [Online]. Available:
http://arxiv.org/abs/1811.01190
[8] M. Parekh and G. Kulkarni, “A Survey on ‘Malware Analysis Techniques, its Detection and
Mitigation.,’” vol. 08, no. 08, 2021.
[9] W. Aman, “A Framework for Analysis and Comparison of Dynamic Malware Analysis Tools,” Int. J.
Netw. Secur. Its Appl., vol. 6, no. 5, pp. 63–74, Sep. 2014, doi: 10.5121/ijnsa.2014.6505.
[10] D. Zeng, C. Noteboom, K. Sutrave, and R. Godasu, “A Meta-Analysis of Evolution of Deep Learning
Research in Medical Image Analysis”.
[11] S. Yoo, S. Kim, S. Kim, and B. B. Kang, “AI-HydRa: Advanced hybrid approach using random forest
and deep learning for malware classification,” Inf. Sci., vol. 546, pp. 420–435, Feb. 2021, doi:
10.1016/j.ins.2020.08.082.
[12] B. Sun, Q. Li, Y. Guo, Q. Wen, X. Lin, and W. Liu, “Malware family classification method based on
static feature extraction,” in 2017 3rd IEEE International Conference on Computer and
Communications (ICCC), Chengdu, Dec. 2017, pp. 507–513. doi:
10.1109/CompComm.2017.8322598.

124
[13] S.-H. Zhang, C.-C. Kuo, and C.-S. Yang, “Static PE Malware Type Classification Using Machine
Learning Techniques,” in 2019 International Conference on Intelligent Computing and its Emerging
Applications (ICEA), Tainan, Taiwan, Aug. 2019, pp. 81–86. doi: 10.1109/ICEA.2019.8858297.
[14] W. S. Noble, “What is a support vector machine?,” Nat. Biotechnol., vol. 24, no. 12, pp. 1565–1567,
Dec. 2006, doi: 10.1038/nbt1206-1565.
[15] T.-Y. Wang and C.-H. Wu, “Detection of packed executables using support vector machines,” in 2011
International Conference on Machine Learning and Cybernetics, Jul. 2011, vol. 2, pp. 717–722. doi:
10.1109/ICMLC.2011.6016774.
[16] C.-W. Chen, C.-H. Su, K.-W. Lee, and P.-H. Bair, “Malware Family Classification using Active
Learning by Learning,” in 2020 22nd International Conference on Advanced Communication
Technology (ICACT), Feb. 2020, pp. 590–595. doi: 10.23919/ICACT48636.2020.9061419.
[17] G. W. Lindsay, “Convolutional Neural Networks as a Model of the Visual System: Past, Present, and
Future,” J. Cogn. Neurosci., vol. 33, no. 10, pp. 2017–2031, Sep. 2021, doi: 10.1162/jocn_a_01544.
[18] A. Fathurrahman, A. Bejo, and I. Ardiyanto, “Lightweight Convolution Neural Network for Image-
Based Malware Classification on Embedded Systems,” in 2021 International Seminar on Machine
Learning, Optimization, and Data Science (ISMODE), Jan. 2022, pp. 12–16. doi:
10.1109/ISMODE53584.2022.9743111.
[19] D. Kornish, J. Geary, V. Sansing, S. Ezekiel, L. Pearlstein, and L. Njilla, “Malware Classification
using Deep Convolutional Neural Networks,” in 2018 IEEE Applied Imagery Pattern Recognition
Workshop (AIPR), Oct. 2018, pp. 1–6. doi: 10.1109/AIPR.2018.8707429.
[20] R. M. Schmidt, “Recurrent Neural Networks (RNNs): A gentle Introduction and Overview.” arXiv,
Nov. 23, 2019. Accessed: Dec. 31, 2022. [Online]. Available: http://arxiv.org/abs/1912.05911
[21] European Union Agency for Cybersecurity., ENISA threat landscape 2022: July 2021 to July 2022.
LU: Publications Office, 2022. Accessed: Jan. 11, 2023. [Online]. Available:
https://data.europa.eu/doi/10.2824/764318

MACHINE LEARNING APPLICATIONS IN MALWARE CLASSIFICATION: A METAANALYSIS LITERATURE REVIEW

Recommended

More Related Content

Similar to MACHINE LEARNING APPLICATIONS IN MALWARE CLASSIFICATION: A METAANALYSIS LITERATURE REVIEW (20)

Recently uploaded (20)

MACHINE LEARNING APPLICATIONS IN MALWARE CLASSIFICATION: A METAANALYSIS LITERATURE REVIEW