A New Machine Learning Model for Perimeter Intrusion Detection
Author Information
Author(s): Pitafi Shahneela, Anwar Toni, Widia I Dewa Made, Sharif Zubair, Yimwadsana Boonsit
Primary Institution: Computer & Information Sciences Department (CISD), Universiti Teknologi PETRONAS, Malaysia
Hypothesis
Can a novel machine learning model improve perimeter intrusion detection accuracy?
Conclusion
The proposed model significantly enhances the accuracy of perimeter intrusion detection systems.
Supporting Evidence
- The proposed model achieved a silhouette score of 0.86, outperforming existing techniques.
- The enhanced DBSCAN algorithm improved detection accuracy by effectively distinguishing between intrusion types.
- The study utilized a dataset of 7,962 images across seven classes of intrusion activities.
Takeaway
This study created a smart system that can better detect when someone tries to break into a place by looking at pictures.
Methodology
The study used a pre-trained InceptionV3 model for feature extraction, followed by t-SNE for dimensionality reduction and an enhanced DBSCAN algorithm for clustering.
Limitations
The study only considered six types of intrusion activities and did not address intrusions from above the perimeter fence.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website