Abstract— This research outlines the cloud data security when using machine learning techniques – Random Forest, Deep Neural Networks, and Q-Learning to prevent unauthorized data transfers and leaks. The major findings point to the fact that DNN showed a higher level of prediction capabilities in comparison with Random Forest – 95% of overall accuracy as opposed to 92%. It is particularly important to consider AUC-ROC of Random Forest, which is 0.96, making it the most reliable model. However, Q-Learning appears to be less accurate with 88% yet more effective when it comes to a cumulative reward and a policy optimization – features that are vital for a changing environment of cloud servers. The findings of the research make a significant contribution to the field of cloud data security, showing the effectiveness of advanced machine learning models in terms of identifying and mitigating security breaches. This, in turn, creates the opportunity of implementation of these techniques into security frameworks for enhancing the resilience and efficiency of the latter. The recommendations for further research lie in the area of hybrid models creation, in particular, the models that would be able to utilize the positive sides of all three techniques. Moreover, the results would be more generalized with a significantly larger dataset comprising diverse cloud environments and threat situations. Finally, the investigation of the models described in a real-time setting and large cloud-based
systems may be suggested for further research, given that these characteristics are essential for an effective practical deployment helping resist emerging threats.
Keywords- Cloud Data Security, Machine Learning, Random Forest, Deep Neural Networks, Q-Learning
Bharda Priya Dutt, M. Prasad, D. K. S. Rao· International Scientific Jou...· 0 citations
Rapidly increasing numbers of devices connected to the Internet of Things have greatly changed how modern digital ecosystems operate by facilitating interoperability among industries such as healthcare, intelligent transportation systems, smart city technology and industrial automation. The increasing number of Internet connected devices has created new vulnerabilities for cyber-attacks on these devices, such as DDoS, Botnets, unauthorized access to devices and data breaches. Traditional IoT protection approaches generally fail to address these types of threats due to their limited resources, heterogeneity, and constant changes. In this regard, artificial intelligence has been proposed as a viable method to enhance cybersecurity for IoTs via AI-based systems capable of intelligent threat detection, adaptive learning, and real-time responses.
To provide a clear understanding of AI-based cybersecurity methods for IoTs, this paper presents a systematic literature review (SLR) using a structured review process that follows the PRISMA guidelines. A total of 3072 original research papers on AI-based cybersecurity methods for IoTs published in top indexing databases were obtained. Each selected paper was analysed using a systematic screening process. The studies were grouped into six distinct theme areas, including machine learning-based intrusion detection, deep learning-based anomaly detection, hybrid/ensemble security models, federated learning frameworks, explainable AI, and blockchain-enabled IoT security mechanisms.
Machine learning algorithms demonstrated high computational efficiency in detecting intrusions but lacked the flexibility to adapt to new threats. On the other hand, deep learning algorithms demonstrated a high intrusion detection capability but at a high computational cost. Hybrid/federated learning algorithms represent an emerging class that can deliver both the required accuracy and scalability while preserving user privacy. While several important advancements exist in AI-based cybersecurity methods for IoTs, many additional areas will require significant work before widespread adoption. Some examples include continued reliance on outdated benchmarks to evaluate algorithmic performance; the absence of validation testing for algorithms running in real-time environments; difficulty interpreting results generated by black-box AI-based algorithms; and a lack of collaboration among researchers studying different layers of the edge-cloud architecture.
As a result of synthesising the current state of the art in AI-based cybersecurity methods for IoTs, this paper proposes a conceptual model of an AI-based adaptive cybersecurity system for IoTs, designed to utilise edge intelligence, federated learning, and continuous feedback mechanisms to detect and respond dynamically to potential threats. This paper makes contributions to the current body of knowledge regarding the rapidly changing landscape of IoT cybersecurity by providing an organised, analytical summary of the current state of the art in AI-based cybersecurity methods for IoTs; defining specific research gaps; and providing recommendations for future research to develop scalable, understandable, and adaptable security systems for IoT environments.
M. Prasad, D. Kumar, Debasish Paul· Journal of Intelligent Decis...· 0 citations