{
  "abstract": "We develop a class of general diffusion-based stochastic particle optimisation methods for loss functions with intractable gradients. Specifically, we consider problems in which the loss gradient is an integral with respect to a parameter-dependent distribution, a structure that includes training generative models, fine-tuning, and learning latent-variable models. We introduce mean-field dynamics and its interacting-particle approximations, which contain several existing algorithms as special cases and provide a route to constructing new methods. Under well-posedness and joint contractivity assumptions, we prove exponential convergence and show that the continuous-time particle system admits a non-asymptotic error bound. We illustrate our framework by developing higher-order Langevin and momentum optimisers and evaluating them on maximum marginal-likelihood estimation and energy-based-model training.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2608.02844",
  "author_total": 2,
  "authors": [
    {
      "name": "Jiechen Jackie Zhang"
    },
    {
      "name": "O. Deniz Akyildiz"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/22mrf9ewbg3tf.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2608.02844",
  "fulltext_url": null,
  "id": "22mrf9ewbg3tf",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2608.02844",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.CO"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Stochastic Gradient Optimization Techniques"
    ]
  ],
  "openalex": "W7172516808",
  "pdf_url": "https://arxiv.org/pdf/2608.02844",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-09",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/22mrf9ewbg3tf.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:28:46Z",
      "native_id": "2608.02844",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2608.02844"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.CO",
      "subfield": "statistics"
    }
  ],
  "title": "Particle-based Generalised Stochastic Optimisation",
  "title_text": "Particle-based Generalised Stochastic Optimisation",
  "type": "preprint",
  "updated_at": "2026-10-30T12:28:46Z",
  "url": "https://research.draconitic.com/paper/22mrf9ewbg3tf",
  "venue": "arXiv",
  "versions": []
}