The development of effective monitoring software agents, essential components of modern multi-agent systems (MAS), increasingly relies on sophisticated model structures such as multi-layer machine learning ensembles. However, the growing complexity of these architectures presents significant challenges in ensuring the reliability, auditability, and, most critically, the reproducibility of experimental results. Addressing this challenge, this paper proposes a declarative approach centered on a newly developed domain-specific language (DSL). As a research method, this language provides a structured, human-readable format for describing the entire model construction process. The DSL specification covers not only data preparation and hyperparameter optimization but also the intricate configuration of multi-layer ensembles, including advanced mechanisms like recirculation and data homogeneity improvement through clustering. A software system was developed to interpret this DSL, thereby automating the complex training process. A key feature of this process is the automatic generation of a self-contained, reproducible bundle. This bundle includes not only the serialized models and preprocessing steps but also all associated configurations, performance metrics, and detailed provenance data, ensuring no implicit dependencies remain. The main results demonstrate that this declarative approach effectively tackles the complexity of managing advanced experiments, ensures the integrity of the created models, and guarantees their full reproducibility. It was also found that formalizing the experimental setup in a DSL provides a robust and objective framework for comparing different, often heterogeneous, model architectures. In conclusion, the proposed DSL-oriented approach creates a reliable and auditable foundation for developing and validating effective software agents. This work bridges the critical gap between algorithmic research and the practical need for trustworthy and deployable machine learning systems in real-world applications.
[1] Wooldridge, M. (2009). An Introduction to MultiAgent Systems. John Wiley & Sons..
[2] Голуб, С. (2007). Багаторівневе моделювання в технологіях моніторингу оточуючого середовища. Черкаси: Вид. Від. ЧНУ Імені Богдана Хмельницького, 220.
[3] Holub, S. V., Ostapiuk, V. V.(2025). Machine learning of multilayer models of a monitoring software agent. Mathematical Machines and Systems, 2, 76–96. https://doi.org/10.34121/1028-9763-2025-2-76-95.
[4] Hutson, M. (2018). Artificial intelligence faces reproducibility crisis. Science, 359(6377), 725–726. https://doi.org/10.1126/ science.359.6377.725.
[5] Zaharia, M. A., Chen, A., Davidson, A., Ghodsi, A., Hong, S. A., Konwinski, A., Murching, S., Nykodym, T., Ogilvie, P., Parkhe, M., Xie, F., & Zumar, C. (2018). Accelerating the Machine Learning Lifecycle with MLflow. IEEE Data Eng. Bull., 41, 39–45.
[6] The Kubeflow Authors. (2025). Kubeflow Pipelines. https://github.com/kubeflow/pipelines
[7] Feurer, M., Klein, A., Eggensperger, K., Springenberg, J. T., Blum, M., & Hutter, F. (2015, December 7). Efficient and Robust Automated Machine Learning. Neural Information Processing Systems. https://www.semanticscholar.org/ paper/Efficient-and-Robust-Automated-Machine-Learning-Feurer-Klein/775a4e375cc79b53b94e37fa3eedff481823e4a6
[8] Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., Kudlur, M., Levenberg, J., Monga, R., Moore, S., Murray, D. G., Steiner, B., Tucker, P., Vasudevan, V., Warden, P., … Zheng, X. (2016). TensorFlow: A system for large-scale machine learning (No. arXiv:1605.08695). arXiv. https://doi.org/10.48550/ arXiv. 1605.08695.
[9] Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Kopf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., … Chintala, S. (2019). PyTorch: An Imperative Style, High-Performance Deep Learning Library. Advances in Neural Information Processing Systems, 32. https://papers.nips.cc/paper_files/ paper/2019/hash/ bdbca288fee7f92f2bfa9f7012727740-Abstract.html
[10] Wolpert, D. H. (1992). Stacked generalization. Neural Networks, 5(2), 241–259. https://doi.org/10.1016/S0893-6080(05)80023-1.
[11] Akiba, T., Sano, S., Yanase, T., Ohta, T., & Koyama, M. (2019). Optuna: A Next-generation Hyperparameter Optimization Framework (No. arXiv:1907.10902). arXiv. http://arxiv. org/abs/1907.10902