國立虎尾科技大學 |

Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks./
作者:	Smith, Daryle Q., Jr.
面頁冊數:	1 online resource (135 pages)
附註:	Source: Dissertations Abstracts International, Volume: 85-07, Section: B.
Contained By:	Dissertations Abstracts International85-07B.
標題:	Information technology. -
電子資源:	click for full text (PQDT)
ISBN:	9798381403756

Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks.
Smith, Daryle Q., Jr.

Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks. - 1 online resource (135 pages)

Source: Dissertations Abstracts International, Volume: 85-07, Section: B.

Thesis (Ph.D.)--North Carolina Agricultural and Technical State University, 2023.

Includes bibliographical references

This research focuses on the detection of malware in sample datasets using machine learning algorithms. As malware becomes more advanced against detection in the cybersecurity sector, researchers have looked for other methods to help with mitigation and exposure. The use of machine learning algorithms has progressed in recent years and has also proven to be effective in the discovery of malware. This research creates a model using three unsupervised learning techniques (K-Means, DBSCAN, GMM) and seven supervised learning techniques (Decision Tree, Random Forest, Ada Boost, KNeighbors, Stochastic Gradient Descent, Extra Trees, and Gaussian Naive Bayes) to detect malware within two public datasets, Malware-Exploratory and CIC-MalMem-2022, and two created datasets SMITH and GAN. The research is further enhanced to include feature selection using Pearson correlation coefficient and the genetic algorithm, followed by deep learning using Long Short Term Memory and explainable AI.The main contribution of the unsupervised learning algorithms is to test against anomalies within the datasets. Results revealed that the datasets have consistencies between the three and that no abnormalities existed. For the supervised learning algorithms results show that the model does very well in detecting malware. For the Malware-Exploratory dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 89.67%, Pearson at 84.92%, and GA at 86.25%. For the CIC-MalMem2022 dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 99.72%, Pearson at 99.25%, and GA at 99.75%. For the SMITH dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 95.19%, Pearson at 97.10%, and GA at 94.98%. For the GAN-based dataset, accuracy averaged scores across each algorithm are as follows: Raw data and Pearson at 84.54%, and GA at 83.91%. The GAN-based dataset reported the lowest accuracy using the traditional machine learning algorithms. However, LSTM eliminated the vanishing gradient issue that the dataset was faced with, accurately detecting up to 98.00% at 20 epochs. Th results from explainable AI proved that non meaningful features within the dataset can cause the algorithms to weigh heavily against noise.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798381403756Subjects--Topical Terms:

559429
Information technology.
Subjects--Index Terms:

Explainable AIIndex Terms--Genre/Form:

554714
Electronic books.

Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks.
LDR:03720ntm a22003857 4500 001 1143826
005 20240517105003.5
006 m o d
007 cr mn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798381403756
035 $a (MiAaPQ)AAI30571025
035 $a AAI30571025
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Smith, Daryle Q., Jr. $3 1468626
245 1 0 $a Exploring Machine Learning for Malware Detection With Feature Selection, Explainable AI, and Generative Adversarial Networks.
264 0 $c 2023
300 $a 1 online resource (135 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-07, Section: B.
500 $a Advisor: Roy, Kaushik.
502 $a Thesis (Ph.D.)--North Carolina Agricultural and Technical State University, 2023.
504 $a Includes bibliographical references
520 $a This research focuses on the detection of malware in sample datasets using machine learning algorithms. As malware becomes more advanced against detection in the cybersecurity sector, researchers have looked for other methods to help with mitigation and exposure. The use of machine learning algorithms has progressed in recent years and has also proven to be effective in the discovery of malware. This research creates a model using three unsupervised learning techniques (K-Means, DBSCAN, GMM) and seven supervised learning techniques (Decision Tree, Random Forest, Ada Boost, KNeighbors, Stochastic Gradient Descent, Extra Trees, and Gaussian Naive Bayes) to detect malware within two public datasets, Malware-Exploratory and CIC-MalMem-2022, and two created datasets SMITH and GAN. The research is further enhanced to include feature selection using Pearson correlation coefficient and the genetic algorithm, followed by deep learning using Long Short Term Memory and explainable AI.The main contribution of the unsupervised learning algorithms is to test against anomalies within the datasets. Results revealed that the datasets have consistencies between the three and that no abnormalities existed. For the supervised learning algorithms results show that the model does very well in detecting malware. For the Malware-Exploratory dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 89.67%, Pearson at 84.92%, and GA at 86.25%. For the CIC-MalMem2022 dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 99.72%, Pearson at 99.25%, and GA at 99.75%. For the SMITH dataset, accuracy averaged scores across each algorithm are as follows: Raw data at 95.19%, Pearson at 97.10%, and GA at 94.98%. For the GAN-based dataset, accuracy averaged scores across each algorithm are as follows: Raw data and Pearson at 84.54%, and GA at 83.91%. The GAN-based dataset reported the lowest accuracy using the traditional machine learning algorithms. However, LSTM eliminated the vanishing gradient issue that the dataset was faced with, accurately detecting up to 98.00% at 20 epochs. Th results from explainable AI proved that non meaningful features within the dataset can cause the algorithms to weigh heavily against noise.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Information technology. $3 559429
650 4 $a Computer science. $3 573171
653 $a Explainable AI
653 $a Generative adversarial networks
653 $a Machine learning
653 $a Malware detection
655 7 $a Electronic books. $2 local $3 554714
690 $a 0984
690 $a 0489
690 $a 0800
710 2 $a North Carolina Agricultural and Technical State University. $b Computer Science. $3 1190751
710 2 $a ProQuest Information and Learning Co. $3 1178819
773 0 $t Dissertations Abstracts International $g 85-07B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30571025 $z click for full text (PQDT)