# UNSW-NB15 Computer Security Dataset: Analysis through Visualization

arXiv: https://arxiv.org/abs/2101.05067 · HTML: https://mlsanity.com/paper/2101.05067/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Jan 2021.

## Built on by (most cited first, 7 of 7)

- [Attraction-Repulsion Swarming: A Generalized Framework of t-SNE via Force Normalization and Tunable Interactions](https://mlsanity.com/paper/2411.10617/) (2024)
- [NBcoded: network attack classifiers based on Encoder and Naive Bayes model for resource limited devices](https://mlsanity.com/paper/2109.07273/) (2021)
- [Ensemble Classifier Design Tuned to Dataset Characteristics for Network Intrusion Detection](https://mlsanity.com/paper/2205.06177/) (2022)
- [Sequential Binary Classification for Intrusion Detection](https://mlsanity.com/paper/2406.06099/) (2024)
- [A Comparative Analysis of Machine Learning Models for Intrusion Detection in Intelligent Transport Systems](https://mlsanity.com/paper/2605.00279/) (2026)
- [Provenance Guided Incremental Learning Under Evolving Concept Definitions](https://mlsanity.com/paper/2608.23893/) (2026)
- [Candidate Comparability Before Promotion: Conditional Validation in Adaptive Network Intrusion Detection](https://mlsanity.com/paper/2609.04388/) (2026)

## Abstract (arXiv)

This paper presents a visual analysis of the UNSW-NB25 computer network security or intrusion detection dataset in order to detect any issues inherent to this dataset which may require researchers to address before employing this dataset for data-driven model development such as a machine learning classifier. A number of data preprocessing algorithms are applied on the raw data to address common issues such as elimination of redundant features, conversion of nominal features into numerical format and scaling. PCA, t-SNE and K-means clustering algorithms are employed for developing the graphs and plots for visualization. Consequent analysis through visualization identified and illustrated two major problems as class imbalance and class overlap for this dataset. In conclusion, it is necessary to address these two problems of class imbalance and class overlap prior to employing this dataset for any classifier model development.