Skripsi

IMPLEMENTATION AND ANALYSIS OF THE STATE-ACTION-REWARD-STATE-ACTION (SARSA) ALGORITHM AND DEEP Q-NETWORK FOR MALWARE CLASSIFICATION

---

Flooding is one Malware is one of the major threats to cybersecurity and continues to evolve, making it increasingly difficult to detect using conventional methods. This study aims to implement and analyze the State-Action-Reward-State-Action (SARSA) and Deep Q-Network (DQN) algorithms for binary malware classification based on static features extracted from the dataset. Data preprocessing, feature selection, and dataset partitioning into training, validation, and testing sets were conducted prior to model development. SARSA utilized discrete state representations generated through K-Means clustering, while DQN employed a neural network to estimate Q-values directly from numerical features. The results show that both algorithms were capable of classifying malware effectively. SARSA achieved an average testing accuracy of 91.50% with an average F1-score of 90.99%, while DQN achieved an average testing accuracy of 88.45% with an average F1-score of 87.54%. These findings indicate that SARSA outperformed DQN and provided more stable classification performance on the dataset used in this study. In conclusion, both SARSA and DQN were successfully implemented for malware classification. However, SARSA demonstrated better classification performance and generalization capability than DQN.

---

Availability

Detail Information

Series Title

-

Call Number

T2011382026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xvi, 189 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

005.807

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Keamanan Data
Prodi Sistem Komputer

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment