{
  "abstract": "State-space models (SSMs) provide a general framework for representing latent states, temporal transitions, and observations in dynamical systems. However, conventional state-space modeling typically assumes that the model structure, state representation, transition mechanism, and observation mechanism are predefined or adapted within a specified model family. This paper proposes Meta-State-Space Models (Meta-SSMs) as a conceptual framework in which a higher-level mechanism dynamically constructs, selects, configures, and restructures state-space models according to observed data, task requirements, environmental conditions, uncertainty, and computational constraints. Rather than merely estimating the latent state of a system, the meta-mechanism operates on the mechanisms that define how latent states are represented, updated, and translated into observations. We formalize the framework using a meta-state that summarizes the current configuration of the underlying state-space model, a candidate-model generator, a model-selection mechanism, and an adaptive update rule. The framework distinguishes parameter adaptation, mechanism adaptation, structural adaptation, and model-family generation. We introduce a multi-objective formulation balancing predictive performance, uncertainty, computational cost, adaptation cost, and structural complexity. We further discuss hierarchical state-space modeling, recursive meta-adaptation, stability, identifiability, and evaluation protocols. Importantly, existing research on meta-learning for neural state-space models and meta-state-space representations of stochastic dynamical systems establishes substantial conceptual overlap with parts of this proposal. Accordingly, the contribution claimed here is a proposed unifying formulation rather than a demonstrated first invention of meta-state-space modeling. No empirical superiority is claimed; experiments are required to determine whether the proposed framework provides measurable benefits over existing approaches.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": null,
  "author_total": 1,
  "authors": [
    {
      "name": "Jinchen Zhang",
      "orcid": "0009-0005-1860-0009"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/8kntkay2hx1tk.bib",
  "cited_by": 0,
  "doi": "10.5281/zenodo.23530028",
  "fulltext_url": null,
  "id": "8kntkay2hx1tk",
  "issns": [],
  "landing_url": "https://zenodo.org/records/23530028",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [],
  "openalex": null,
  "pdf_url": "https://zenodo.org/api/records/23530028/files/Meta-State-Space%20Models.pdf/content",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-21",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/8kntkay2hx1tk.ris",
  "sources": [
    {
      "first_seen": "2026-11-04T12:21:15Z",
      "native_id": "23530028",
      "source": "zenodo",
      "url": "https://zenodo.org/records/23530028"
    }
  ],
  "subjects": [],
  "title": "Meta-State-Space Models: A General Framework for Generating, Adapting, and Restructuring State-Space Models Through a Higher-Level Dynamic Mechanism",
  "title_text": "Meta-State-Space Models: A General Framework for Generating, Adapting, and Restructuring State-Space Models Through a Higher-Level Dynamic Mechanism",
  "type": "preprint",
  "updated_at": "2026-11-04T12:21:15Z",
  "url": "https://research.draconitic.com/paper/8kntkay2hx1tk",
  "venue": "Zenodo",
  "versions": []
}