AI2030 Responsible AI Tools Marketplace

Key Takeaways

Explainability matters for IoT security: The framework incorporates LIME and other Explainable AI techniques to make security decisions transparent, addressing a key concern in responsible AI—users and administrators can understand why the system flagged a potential threat rather than treating it as a "black box."
Lightweight design enables broader deployment: By using edge-optimized quantization and pruning, the model can run directly on IoT devices rather than requiring constant cloud connectivity, reducing privacy risks and making security accessible to smaller organizations that may lack extensive infrastructure.
High accuracy with practical robustness: The system achieved 98% accuracy on real-world IoT attack datasets (CIC-IoV-2024 and CIC-APT-IIoT-2024) while remaining robust against noisy data, suggesting it can reliably protect networks in messy, real-world conditions rather than just controlled lab environments.
Staged learning reduces bias and improves fairness: The curriculum learning approach iteratively removes low-relevance features and refines the model in stages, which can help prevent the system from over-relying on potentially biased patterns and improve performance across different types of IoT networks and attack scenarios.
Ensemble approach increases reliability: Combining multiple algorithms (Random Forest, XGBoost, and neural networks) rather than relying on a single model reduces the risk of catastrophic failures and makes the security system more trustworthy for critical infrastructure protection.

Abstract

To address the critical need for secure IoT networks, this study presents a scalable and lightweight curriculum learning framework enhanced with Explainable AI (XAI) techniques, including LIME, to ensure transparency and adaptability. The proposed model employs novel neural network architecture utilized at every stage of Curriculum Learning to efficiently capture and focus on both short- and long-term temporal dependencies, improve learning stability, and enhance accuracy while remaining lightweight and robust against noise in sequential IoT data. Robustness is achieved through staged learning, where the model iteratively refines itself by removing low-relevance features and optimizing performance. The workflow includes edge-optimized quantization and pruning to ensure portability that could easily be deployed in the edge-IoT devices. An ensemble model incorporating Random Forest, XGBoost, and the staged learning base further enhances generalization. Experimental results demonstrate 98% accuracy on CIC-IoV-2024 and CIC-APT-IIoT-2024 datasets and 97% on EDGE-IIoT, establishing this framework as a robust, transparent, and high-performance solution for IoT network security.

2 tables, 5 figures

Key Takeaways

Explainability matters for IoT security: The framework incorporates LIME and other Explainable AI techniques to make security decisions transparent, addressing a key concern in responsible AI—users and administrators can understand why the system flagged a potential threat rather than treating it as a "black box."
Lightweight design enables broader deployment: By using edge-optimized quantization and pruning, the model can run directly on IoT devices rather than requiring constant cloud connectivity, reducing privacy risks and making security accessible to smaller organizations that may lack extensive infrastructure.
High accuracy with practical robustness: The system achieved 98% accuracy on real-world IoT attack datasets (CIC-IoV-2024 and CIC-APT-IIoT-2024) while remaining robust against noisy data, suggesting it can reliably protect networks in messy, real-world conditions rather than just controlled lab environments.
Staged learning reduces bias and improves fairness: The curriculum learning approach iteratively removes low-relevance features and refines the model in stages, which can help prevent the system from over-relying on potentially biased patterns and improve performance across different types of IoT networks and attack scenarios.
Ensemble approach increases reliability: Combining multiple algorithms (Random Forest, XGBoost, and neural networks) rather than relying on a single model reduces the risk of catastrophic failures and makes the security system more trustworthy for critical infrastructure protection.

Abstract

2 tables, 5 figures

Enhancing IoT Network Security through Adaptive Curriculum Learning and XAI

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links

Enhancing IoT Network Security through Adaptive Curriculum Learning and XAI

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links