Liquid NeurIoTic: Liquid Neural Network-Based Intrusion Detection System for Internet of Things in Healthcare Networks
Open Access DepositedOver 249.09 million individuals were affected by healthcare cyberattacks between 2005 and 2019. Alarmingly, a 95% probability exists that distributed denial-of-service (DDoS) attacks on the Internet of Things in Healthcare (IoT-H) devices pose a life-threatening risk. Despite notable progress in the field, current solutions for neural networks (NN) with IoT-H networks are flawed by issues regarding scalability and IoT device resource management. As a viable alternative to conventional NNs in detecting intrusions in IoT-H networks, this praxis introduces Liquid NeurIoTic, a liquid neural network (LNN)-based intrusion detection system for IoT-H networks. This praxis used the CICIoT2023 dataset to train and validate the LNN studied by identifying and categorizing distributed denial-of-service, denial-of-service, Mirai, spoofing, and reconnaissance attacks. This praxis also provided a novel technique to convert network traffic features into cyclical waveforms to enable time-series representations for enhanced integration into an LNN model. As a result, Liquid NeurIoTic reduced the number of neurons by as much as 40% and the number of epochs by as much as 64% compared to the best-performing artificial and recurrent neural networks. It also achieved a 61.81% reduction in training time. Additionally, Liquid NeurIoTic stood out in the broader NN research landscape because it used fewer neurons, epochs, and parameters than models in the existing literature and exceeded performance in accuracy, recall, precision, and F1-score metrics.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
Kramarczyk_gwu_0075A_16984.pdf | 2025-04-09 | Open Access |
|