{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T05:16:50Z","timestamp":1783487810727,"version":"3.55.0"},"reference-count":18,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T00:00:00Z","timestamp":1783468800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100006769","name":"Russian Science Foundation","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100006769","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Big Data"],"abstract":"<jats:p>\n                    Demographic systems face unprecedented challenges from simultaneous crises. Conventional statistical demography techniques and agent\u2013based models often struggle to capture nonlinear inter\u2013regional interactions during periods of severe socio\u2013economic disruption. To address this, we propose\n                    <jats:italic>MADDPG\u2013EVO\u2013DGM<\/jats:italic>\n                    , a hybrid algorithm that integrates multi\u2013agent deep reinforcement learning with evolutionary optimisation and meta\u2013learning principles to model regional demographic processes under multiple crisis scenarios. Each region is treated as an autonomous agent learning to steer demographic policy levers, while periodic evolutionary \u201cboosters\u201d overcome local optima via population\u2013based perturbations of actor network parameters. Additionally, a Darwin\u2013G\u00f6del Machine\u2013inspired meta\u2013learning mechanism adapts the booster triggers, enabling self\u2013improvement in the learning process. We evaluate MADDPG\u2013EVO\u2013DGM on a simulation environment calibrated with real demographic data for eight federal regions of the Russian Federation over the period 2000\u20132024 and subject to ten concurrent crisis scenarios (e.g., pandemic, geopolitical conflict, economic collapse). Experiments demonstrate significantly faster convergence and improved performance over a baseline MADDPG: the hybrid approach achieves a higher final average reward (252.57 vs. 243.07) and 3.4 \u00d7 lower convergence variance (\u03c3 = 0.24 vs. 0.80), indicating more reliable training. It also exhibits qualitative performance jumps of +68% during evolutionary phases and maintains 35%\u201345% greater resilience under crisis shocks compared to the baseline. To our knowledge, this is the first application of multi\u2013agent reinforcement learning to large\u2013scale demographic modeling under crises, opening new possibilities for evidence\u2013based, crisis\u2013resilient population policy design. Code, data, and logs are provided to ensure reproducibility.\n                  <\/jats:p>","DOI":"10.3389\/fdata.2026.1842233","type":"journal-article","created":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T04:51:30Z","timestamp":1783486290000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Evolutionary multi-agent reinforcement learning for crisis-aware demographic policy optimization"],"prefix":"10.3389","volume":"9","author":[{"given":"Anton V.","family":"Dozhdikov","sequence":"first","affiliation":[{"name":"Federal Research Sociological Centre of the Russian Academy of Sciences","place":["Moscow, Russia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Arseniy M.","family":"Sitkovskiy","sequence":"additional","affiliation":[{"name":"Federal Research Sociological Centre of the Russian Academy of Sciences","place":["Moscow, Russia"]},{"name":"Peoples Friendship University of Russia (RUDN University)","place":["Moscow, Russia"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,7,8]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-7908-2715-6","author":"Billari","year":"2003","journal-title":"Agent-Based Computational Demography: Using Simulation to Improve Our Understanding of Demographic Behaviour"},{"key":"B2","doi-asserted-by":"publisher","first-page":"9187","DOI":"10.1073\/pnas.1722364115","article-title":"Forecast accuracy hardly improves with method complexity when completing cohort fertility forecasts","volume":"115","author":"Bohk-Ewald","year":"2018","journal-title":"Proc. Nat. Acad. Sci"},{"key":"B3","unstructured":"Regions of Russia: Socio-Economic Indicators"},{"key":"B4","unstructured":"Unified Interdepartmental Statistical Information System (EMISS)"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v32i1.11794","article-title":"\u201cCounterfactual multi-agent policy gradients,\u201d","author":"Foerster","year":"2018","journal-title":"Proceedings of the AAAI conference on artificial intelligence, AAAI 2018"},{"key":"B6","article-title":"Population based training of neural networks","author":"Jaderberg","year":"2017","journal-title":"arXiv preprint arXiv:1711.09846"},{"key":"B7","first-page":"1196","article-title":"\u201cEvolution-guided policy gradient in reinforcement learning,\u201d","author":"Khadka","year":"2018","journal-title":"Advances in neural information processing systems, NeurIPS 2018"},{"key":"B8","doi-asserted-by":"publisher","first-page":"659","DOI":"10.1080\/01621459.1992.10475265","article-title":"Modeling and forecasting U. S. mortality","volume":"87","author":"Lee","year":"1992","journal-title":"J. Am. Stat. Assoc"},{"key":"B9","article-title":"Multi-agent actor-critic for mixed cooperative-competitive environments","author":"Lowe","year":"2017","journal-title":"arXiv preprint arXiv:1706.02275"},{"key":"B10","first-page":"1535","article-title":"\u201cEvolutionary reinforcement learning for sample-efficient multiagent coordination,\u201d","author":"Majumdar","year":"2020","journal-title":"Proceedings of the international conference on autonomous agents and multiagent systems"},{"key":"B11","article-title":"\u201cCem-rl: combining cross-entropy method and reinforcement learning,\u201d","author":"Pourchot","year":"2019","journal-title":"International conference on learning representations, ICLR 2019"},{"key":"B12","first-page":"4295","article-title":"\u201cQmix: monotonic value function factorisation for deep multi-agent reinforcement learning,\u201d","author":"Rashid","year":"2018","journal-title":"Proceedings of the 35th international conference on machine learning ICML 2018"},{"key":"B13","article-title":"Evolution strategies as a scalable alternative to reinforcement learning","author":"Salimans","year":"2017","journal-title":"arXiv preprint arXiv:1703.03864"},{"key":"B14","doi-asserted-by":"publisher","first-page":"119","DOI":"10.1007\/978-3-540-68677-4_7","article-title":"\u201cG\u00f6del machines: fully self-referential optimal universal self-improvers,\u201d","author":"Schmidhuber","year":"2007","journal-title":"Artificial General Intelligence"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.7148\/2013-0689","article-title":"\u201cSimulating the impact of social care on population dynamics: an agent-based approach,\u201d","author":"Silverman","year":"2013","journal-title":"Proceedings of the European conference on modelling and simulation ECMS 2013"},{"key":"B16","volume-title":"Reinforcement Learning: An Introduction","author":"Sutton","year":"2018"},{"key":"B17","article-title":"Darwin G\u00f6del machine: open-ended evolution of self-improving agents","author":"Zhang","year":"2025","journal-title":"arXiv preprint arXiv:2505.22954"},{"key":"B18","doi-asserted-by":"publisher","first-page":"321","DOI":"10.1007\/978-3-030-60990-0_12","article-title":"\u201cMulti-agent reinforcement learning: a selective overview of theories and algorithms,\u201d","author":"Zhang","year":"2021","journal-title":"Handbook of reinforcement learning and control"}],"container-title":["Frontiers in Big Data"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2026.1842233\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T04:51:41Z","timestamp":1783486301000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2026.1842233\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,8]]},"references-count":18,"alternative-id":["10.3389\/fdata.2026.1842233"],"URL":"https:\/\/doi.org\/10.3389\/fdata.2026.1842233","relation":{},"ISSN":["2624-909X"],"issn-type":[{"value":"2624-909X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,8]]},"article-number":"1842233"}}