MACHINE LEARNING METHODS IN INTELLIGENT «SMART HOME» SYSTEMS: CLASSIFICATION, PROBLEMS, AND PROSPECTS

Максим Олександрович Кушнарьов, Юрій Сергійович Аверін, Ігор Володимирович Шостак

Abstract


The article presents a systematic review of machine learning methods applied in intelligent «Smart Home» systems. A classification of typical tasks is provided, including user activity recognition, resource consumption forecasting, anomaly detection during operation, personalization of household scenarios, comfort management, and health monitoring of residents. Key classes of algorithms are analyzed in detail: classical methods (k-NN, SVM, decision trees, Random Forest), ensemble approaches (Gradient Boosting, XGBoost), deep learning methods (CNN, RNN, LSTM), as well as lightweight models optimized for edge devices (TinyML, TensorFlow Lite). A comparative analysis of their characteristics is conducted, in particular accuracy, resource requirements, adaptability, interpretability, and real-time integration capabilities.

Special attention is given to data organization, sensor networks, and architectural strategies for building «Smart Homes». Key challenges are identified: ensuring privacy and cybersecurity, the lack of high-quality datasets, hardware limitations of devices, as well as issues of standardization and interoperability. Current approaches to overcoming these challenges are analyzed, including data augmentation and transfer learning methods, privacy-enhancing technologies (differential privacy, homomorphic encryption, secure multiparty computation), federated learning, and edge architectures.

Significant attention is devoted to the ethical aspects of AI application, related to model bias, fairness, transparency, accountability, and preservation of user autonomy. Security issues of machine learning models are examined separately: adversarial attacks, data poisoning, and model stealing, as well as respective defense mechanisms — data sanitization, anomaly detection, adversarial training, obfuscation, and watermarking. The role of human-centric approaches and Explainable AI in increasing user trust is highlighted.

Future prospects of the field are outlined, including the spread of Edge AI and TinyML, advancement of federated learning algorithms, implementation of multimodal systems, development of reinforcement learning methods, and the creation of decentralized architectures with local self-learning. The proposed review forms a holistic understanding of the current state and trends of ML use in «Smart Home» systems and defines practical ways to overcome existing problems.


Keywords


Smart Home; machine learning; intelligent systems; personalization; energy efficiency; anomaly detection; federated learning; Edge AI; AI ethics; model security

References


Saleem A. A. Smart Homes Powered by Machine Learning: A Review / A. A. Saleem, M. M. Hassan, I. A. Ali // 2022 International Conference on Computer Science and Software Engineering (CSASE) : proc. of the conf. – 2022. – P. 355–361.

Machine Learning in Smart Buildings: A Review of Methods, Challenges, and Future Trends / F. El Husseini, H. N. Noura, O. Salman, K. Chahine // Applied Sciences. – 2025. – Vol. 15, № 14. – Art. no. 7682.

da Silva D. C. Evaluating Multi-Label Machine Learning Models for Smart Home Environments / D. C. da Silva, D. R. D. Boaventura, M. dos Santos Oliveira // Software: Practice and Experience. – 2025.

Smart Home System: A Comprehensive Review / A. Chakraborty [et al.] // Journal of Electrical and Computer Engineering. – 2023. – Vol. 2023. – P. 1–30.

Latoń D. Applications of Deep Reinforcement Learning for Home Energy Management Systems: A Review / D. Latoń, J. Grela, A. Ożadowicz // Energies. – 2024. – Vol. 17, № 24. – P. 6420.

A Survey on Bias and Fairness in Machine Learning [Electronic resource] / N. Mehrabi [et al.] // arXiv preprint arXiv:1908.09635. – 2022. – Available at: https://arxiv.org/abs/1908.09635.

Detecting and Preventing Data Poisoning Attacks on AI Models / H. Kure, P. Sarkar, A. B. Ndanusa, A. O. Nwajana // 2025 PhotonIcs & Electromagnetics Research Symposium (PIERS) : proc. of the symp., Abu Dhabi, 2025. – 2025.

Dunn C. Robustness Evaluations of Sustainable Machine Learning Models against Data Poisoning Attacks in the Internet of Things / C. Dunn, N. Moustafa, B. Turnbull // Sustainability. – 2020. – Vol. 12, № 16. – Art. no. 6434.

Iot device security risks: a comprehensive overview and mitigation strategies / U. Iwuanyanwu [et al.] // Journal of Technology & Innovation. – 2023. – Vol. 3, № 1. – P. 38–43.

Explainable Artificial Intelligence: A Survey of Needs, Techniques, Applications, and Future Direction [Electronic resource] / M. Mershaa [et al.] // arXiv preprint arXiv:2409.00265. – 2024. – Available at: https://arxiv.org/abs/2409.00265.

A Detailed Survey on Federated Learning Attacks and Defenses / H. S. Sikandar [et al.] // Electronics. – 2023. – Vol. 12, № 2. – Art. no. 260.

Bias in data-driven artificial intelligence systems—An introductory survey / E. Ntoutsi [et al.] // Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery. – 2020. – Vol. 10, № 3.

Iwana B. K. An empirical survey of data augmentation for time series classification with neural networks / B. K. Iwana, S. Uchida // PLOS One. – 2021. – Vol. 16, № 7. – P. e0254841.

Biggio B. Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning / B. Biggio, F. Roli // Pattern Recognition. – 2018. – Vol. 84. – P. 317–331.

Time Series Data Augmentation for Deep Learning: A Survey [Electronic resource] / Q. Wen [et al.] // arXiv preprint arXiv:2002.12478. – 2022. – Available at: https://arxiv.org/abs/2002.12478.

Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations / NIST. – Gaithersburg, MD : National Institute of Standards and Technology, 2023. – (NIST AI 100-2).

The FAIR Guiding Principles for scientific data management and stewardship / M. D. Wilkinson [et al.] // Scientific Data. – 2016. – Vol. 3. – Art. no. 160018.

The Human in Interactive Machine Learning: Analysis and Perspectives for Ambient Intelligence / K. Delcourt, S. Trouilhet, J.-P. Arcangeli, F. Adreit // Journal of Artificial Intelligence Research. – 2024. – Vol. 81.

A Taxonomy of Attacks in Federated Learning / N. Pitropakis [et al.] // Applied Sciences. – 2021. – Vol. 11, № 21. – Art. no. 10300.

Wang Y. Multi-Agent Reinforcement Learning for Smart Community Energy Management / Y. Wang, T. Huang, K. Li // Energies. – 2024. – Vol. 17, № 20. – P. 5211.

Thakur P. Edge AI Enabled IoT Framework for Secure Smart Home Infrastructure / P. Thakur, S. Goel, E. Puthooran // Procedia Computer Science. – 2024. – Vol. 235. – P. 3369–3378.

Reis M. J. C. S. Edge AI for Real-Time Anomaly Detection in Smart Homes / M. J. C. S. Reis, C. Serôdio // Future Internet. – 2025. – Vol. 17, № 4. – Art. no. 179.

Leveraging Large Language Models for Explainable Activity Recognition in Smart Homes: A Critical Evaluation [Electronic resource] / M. Fiori, G. Civitarese, P. Choudhary, C. Bettini // arXiv preprint arXiv:2503.16622. – 2025. – Available at: https://arxiv.org/abs/2503.16622.

Ethical Dilemmas and Privacy Issues in Emerging Technologies: A Review / L. L. Dhirani, N. Mukhtiar, B. S. Chowdhry, T. Newe // Sensors. – 2023. – Vol. 23, № 3. – P. 1151.

Chung J. Ethical Considerations Regarding the Use of Smart Home Technologies for Older Adults: An Integrative Review / J. Chung, G. Demiris, H. J. Thompson // Annual Review of Nursing Research. – 2016. – Vol. 34, № 1. – P. 155–181.

Vilone G. Explainable AI for Smart Buildings: A Survey on Current Trends, Applications, and Challenges / G. Vilone, L. Longo // Buildings. – 2023. – Vol. 13, № 1. – Art. no. 211.

Utilization of 5G Technologies in IoT Applications: A Systematic Literature Review / I. Makhdoom [et al.] // Sensors. – 2023. – Vol. 23, № 8. – P. 3995.

Stojkoska B. L. R. A review of Internet of Things for smart home: Challenges and solutions / B. L. R. Stojkoska, K. V. Trivodaliev // Journal of Cleaner Production. – 2017. – Vol. 140. – P. 1454–1464.

Matter Specification [Electronic resource] / Connectivity Standards Alliance. – 2022. – Available at: https://csa-iot.org/all-solutions/matter/#spec.

A Survey of Human-in-the-loop for Machine Learning [Electronic resource] / Y. Wu [et al.] // arXiv preprint arXiv:2108.00941. – 2022. – Available at: https://arxiv.org/abs/2108.00941.

Xu X. Privacy Risk Assessment of Smart Home System Based on a STPA–FMEA Method / X. Xu, Z. Li, C. Zhang // Sensors. – 2023. – Vol. 23, № 10. – Art. no. 4664.

Angelov P. Human-Centered XAI for Smart Home Applications [Electronic resource] / P. Angelov, E. Soares // arXiv preprint arXiv:2404.16074. – 2024. – Available at: https://arxiv.org/abs/2404.16074.

Matter: The Foundation for Connected Things [Electronic resource] / Connectivity Standards Alliance. – 2022. – Available at: https://csa-iot.org/all-solutions/matter/#main.

Kim T. A Novel Data Sanitization Method Based on Dynamic Dataset Partition and Inspection Against Data Poisoning Attacks / T. Kim, K. Chae // Sensors. – 2024. – Vol. 24, № 2. – Art. no. 374.

Machine Unlearning: A Comprehensive Survey [Electronic resource] / J. Zhang [et al.] // arXiv preprint arXiv:2405.07406. – 2024. – Available at: https://arxiv.org/abs/2405.07406.

Novak P. Multimodal ML in Ambient Intelligence / P. Novak // Computers in Human Behavior. – 2024.

A Survey of Machine Unlearning Techniques / A. Pawlicka, J. Pawlicki, A. Kozik, M. Choraś // Applied Sciences. – 2023. – Vol. 13, № 24. – Art. no. 13180.

Marikyan D. A systematic review of the literature on the smart home and its users: The pillars of adoption and continued use / D. Marikyan, S. Papagiannidis, E. Alamanos // Technological Forecasting and Social Change. – 2019. – Vol. 146. – P. 660–681.

Challenges and Approaches of Reinforcement Learning for Power System Decision and Control: A Review / J. Zhang [et al.] // Renewable and Sustainable Energy Reviews. – 2023. – Vol. 182.

Recent Advancements in Federated Learning: State of the Art, Fundamentals, Principles, IoT Applications and Future Trends / L. U. Khan [et al.] // IEEE Access. – 2021. – Vol. 9. – P. 130095–130129.




DOI: https://doi.org/10.32620/oikit.2025.105.19

Refbacks

  • There are currently no refbacks.