Scribd
Upload
116 views10 pages

Malware

Uploaded by

Manicka Valli N
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
116 views10 pages

Malware

Uploaded by

Manicka Valli N
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 10

IFET College of Engineering (Autonomous

Institution)

MALWARE DETECTION IN
EXECUTABLE FILE & URL
USING ENSEMBLE LEARNING
TECHNIQUE

Presented by Under the guidance of


Manickavalli N D.Jeyakumar AP/CSE
PROBLEM STATEMENT
The increasing threat of malware, particularly through downloaded files
from the internet, poses significant security risks.
Malware can infiltrate systems, steal sensitive information, and compromise
data integrity.
machine learning-based malware detection to provide a robust and scalable
solution to mitigate security risks associated with file downloads​.
OBJECTIVE:

1. Malware Detection – To develop a system that can detect whether a downloaded


file contains malware using machine learning techniques.
2. Feature Analysis – To analyze various file attributes, such as MD5 hash, optional
header size, and load configuration size, to differentiate between benign and malicious files.
3. Model Training – To train multiple machine learning models, including Decision Tree,
Random Forest, XGBoost, and others, using extracted file features.
4.Enhanced Cybersecurity – To provide an efficient, scalable, and high-accuracy solution for
malware detection, minimizing security risks for users downloading files from the internet.
INTRODUCTION:
These malicious programs can steal sensitive data, encrypt files, or
disrupt system operations, making them a major cybersecurity
concern.
Machine learning provides an effective approach to identifying
malicious files by analyzing key file attributes such as MD5 hash,
optional header size, and load configuration size.
Dataset balancing techniques such as oversampling and SMOTE are
employed to improve classification performance.
LITERATURE SURVEY:
S NO TITTLE PUBLIS CONTENT DRAWBACK
H YEAR
1 Combines static analysis techniques
Hybrid Approach for Malware
with deep learning models like CNN Requires a large amount
Detection Using Deep Learning and 2024
and RNN for detecting malware of labeled data
Static Analysis
patterns.
2 AI-driven model for analyzing High false-positive rate;
Cybersecurity Threat Detection in IoT
2024 network traffic in IoT devices to lacks real-time adaptation
Networks Using AI-Based Models
identify potential security threats. to evolving threats​.
3 Employs deep learning to analyze Requires large labeled
Malware Detuning and Opcode
2023 opcode sequences for malware datasets; struggles with
Analysis
detection. obfuscated malware
4
Training complexity is
Deep Reinforcement Learning for Applies reinforcement learning for
2024 high; requires significant
Adaptive Malware Detection adaptive malware detection
computational resources​.

5 A model using Random Forest for


Using Machine Learning and Signature
executable headers and Logistic Limited feature extraction
Matching to Detect Malware, 2021
Regression for URLs to detect techniques.
Malicious URLs, and Viruses
malware.
EXISTING SYSTEM
Traditional malware detection methods primarily rely on signature-based
techniques, where known malware signatures are stored in a database and
matched against incoming files.
Some antivirus programs use heuristic-based detection, which analyzes a
file’s behavior and structure to identify potential threats.
This approach is resource-intensive, time-consuming, and may not detect
stealthy malware designed to evade sandbox detection.
DRAWBACK:
Inability to Detect New Malware.
Resource-Intensive Processes.
Evasion by Advanced Malware.
Dependency on Frequent Updates.
Limited Scalability .
PROPOSED SYSTEM:
Instead of relying on predefined signatures, the proposed system
analyzes file attributes using machine learning algorithms to classify
files as malicious or benign accurately.
Multiple models, including Decision Tree, Random Forest, and
XGBoost, are trained and evaluated based on accuracy, efficiency,
and F1-score to determine the most effective model.
Techniques like oversampling, under sampling are applied to handle
imbalanced datasets, ensuring better classification performance and
scalability for large-scale malware detection​
ADVANTAGE:
Improved Malware Detection Accuracy.
Detection of Zero-Day and Polymorphic Malware.
Reduced False Positives and False Negatives.
Automated and Scalable Approach.
Adaptability and Continuous Learning.
Efficient Handling of Imbalanced Data.

You might also like