1 of 11

SHARK

FIRE

NeuraDefence

Network Flow-Based Intrusion Detection System

Aditya Sharma, Sejal Kaur, Nikhil Yadav, Rajan Raj

​

2 of 11

Problem:

    • Growing cyber threats with increased digital dependency.

Challenges:

    • Traditional IDS struggle to detect unknown attacks.

Proposed Solution:

    • A Machine Learning-based IDS to monitor network flows.
    • Integrates threat intelligence and email security.

​

THE NEED FOR ENHANCED CYBERSECURITY

3 of 11

PROJECT OBJECTIVES

Real-time analysis of network flows.

Classification of safe vs malicious traffic using Random Forest.

User-friendly interface with threat alerts and analytics.

Email attachment classification into benign or malicious.

Integration with external threat intelligence for enhanced detection.

1

2

3

4

5

4 of 11

FRAMEWORK AND WORKFLOW

Key Components:

    • Real-time monitoring (Wireshark integration).
    • Threat intelligence (AbuseIP Database, blacklists).
    • Email threat classification model.
    • Streamlit-based deployment for user interaction.

​

Advantages:

    • Reduced computational overhead, real-time alerts, and threat intelligence updates.

​

NeuraDefence

5 of 11

EXPERIMENTAL RESULTS

Network Flow Detection:

    • Accuracy: 94.2%, Precision: 93.8%, Recall: 92.9%, F1-Score: 93.3%.
    • Detected 96% of DDoS attacks, 92% of port scans.
    • Low false-positive rate (2.3%).

​

Email Classification:

    • Accuracy: 91.8%, Precision: 90.5%, Recall: 89.7%, F1-Score: 90.1%.
    • Detected 94% phishing attempts, 89% spam campaigns.

​

NeuraDefence

6 of 11

FLOWCHART

7 of 11

KEY FEATURES OF THE SYSTEM

Real-Time Monitoring: Live network traffic analysis.

Threat Lookup: Domain/IP reputation scoring.

Analytics Dashboard: Visualizations (pie charts, histograms) of network activity.

Email Security: Quick classification of attachments into malicious or benign.

1

2

3

4

8 of 11

WHY THIS SYSTEM STANDS OUT

High Accuracy: Reliable detection of threats with minimal false positives.

Real-Time Monitoring: Quick detection and response to threats.

Threat Intelligence Integration: Constantly updated with emerging threats.

User-Friendly Interface: Easy-to-understand insights for administrators.

1

2

3

4

Scalability: Suitable for large networks.

4

9 of 11

    • Deep Learning models like LSTM, CNN.
    • Automated threat response (e.g., IP blocking).

​

WHAT’S NEXT?

Technical Enhancements:

    • Support for additional email providers.
    • Blockchain for secure logging.
    • Mobile application for monitoring.

​

Feature Additions:

    • Firewall and cloud security systems.
    • Automated patch management.

​

Integration Possibilities:

10 of 11

FINAL THOUGHTS

Achievements:

    • Integrated network flow and email threat detection.
    • High accuracy and real-time capabilities.
    • User-friendly design with actionable insights.

​

Significance:

    • Demonstrates the power of machine learning in cybersecurity.
    • A scalable, adaptive solution for modern threats.

NeuraDefence

11 of 11

THANK YOU

NuraDefence

for your time and attention