Summary of Multiple-input Auto-encoder Guided Feature Selection For Iot Intrusion Detection Systems, by Phai Vu Dinh et al.
Multiple-Input Auto-Encoder Guided Feature Selection for IoT Intrusion Detection Systems
by Phai Vu Dinh, Diep N. Nguyen, Dinh Thai Hoang, Quang Uy Nguyen, Eryk Dutkiewicz, Son Pham Bao
First submitted to arxiv on: 22 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a novel neural network architecture called Multiple-Input Auto-Encoder (MIAE) for intrusion detection systems (IDSs). MIAE consists of multiple sub-encoders that process inputs from different sources with varying characteristics. The model is trained in an unsupervised learning mode to transform heterogeneous inputs into lower-dimensional representations, enhancing the accuracy of classifiers in distinguishing between normal behavior and various types of attacks. To further reduce redundant features, a feature selection layer is designed and embedded within MIAE, resulting in MIAEFS. This approach enables the selection of informative features from the representation vector. Experimental results on three IDS datasets demonstrate the superior performance of MIAE and MIAEFS compared to other methods. When combined with Random Forest (RF) classifier, MIAE and MIAEFS achieve high accuracy (96.5%) in detecting sophisticated attacks like Slowloris. The models’ average running time is approximately 1.7E-6 seconds, while the model size remains lower than 1 MB. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps create better tools to detect cyberattacks on the internet of things (IoT). Right now, it’s hard to use machine learning models for this because IoT data is so diverse and high-dimensional. The researchers created a new neural network called MIAE that can handle these complexities by transforming different types of input data into a lower-dimensional space. They also added a feature selection layer to remove unimportant features. This combination outperformed other methods in detecting various types of attacks, including sophisticated ones like Slowloris. |
Keywords
* Artificial intelligence * Encoder * Feature selection * Machine learning * Neural network * Random forest * Unsupervised
