{
  "abstract": "Property prediction for molecules and materials is bottlenecked by label scarcity, and in this regime the operative failure mode is not a low mean but an unpredictable tail: individual training runs collapse. The three dominant molecular representations fail in complementary ways—physicochemical descriptors are exact but fixed, graph neural networks learn topology but degenerate under sparse supervision, and language models supply semantic context but cannot compute exact quantities—yet combining them is itself the learning problem, because modality-level scalar gating commits the whole model to one trust weight per source. We introduce TAME, a tri-modal encoder whose element-wise Mixture-of-Experts router assigns an independent modality mixture to every hidden coordinate, stabilized by a balance–entropy regularizer, a closed-form gate initialization (β = log τ) that removes router burn-in, and two-stage self-supervised graph pretraining. On scaffold-split BACE (≈ 1.5k molecules, 100 seeds), the fusion stage improves level and spread together: median ROC-AUC rises from 0.752 to 0.817 and macro-F1 from 0.634 to 0.703, while seed-to-seed variance falls five-fold and a collapse tail reaching ROC-AUC ≈ 0.45 disappears. Because backbone, checkpoint, split, seeds and optimizer recipe are held fixed, the effect is attributable to the fusion stage alone. The router allocates a distinct mixture to each coordinate, which no scalar gate can represent, and the entropy weight moves that allocation continuously between graded mixing and a near-binary regime in which each coordinate is claimed outright by one modality. Gate mass additionally tracks encoder quality without supervision, shifting from text to topology once pretraining lifts the graph embedding out of rank-1 collapse. The same three experts map onto crystal graphs, composition–structure descriptors and literature-mined text, making TAME a transferable template for multimodal representation learning in materials discovery.",
  "abstract_shown": true,
  "affiliations": [
    "Theoretical Chemistry",
    "University of Munich (LMU",
    "Bielefeld University of Applied Sciences (HSBI",
    "Machine Learning Group",
    "TU Berlin",
    "BIFOLD–Berlin Institute for the Foundations of Learning and Data",
    "Korea University",
    "MPI for Informatics",
    "MPI for Solid State Research"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 5,
  "authors": [
    {
      "affs": [
        0,
        1
      ],
      "name": "Robert P. Schiller",
      "orcid": "0009-0004-5793-678X"
    },
    {
      "affs": [
        2
      ],
      "name": "Christoph Weisser"
    },
    {
      "affs": [
        3,
        4,
        5,
        6,
        7
      ],
      "name": "Klaus-Robert Müller"
    },
    {
      "affs": [
        0,
        1,
        8
      ],
      "name": "Christian Ochsenfeld"
    },
    {
      "affs": [
        3,
        4,
        5
      ],
      "name": "Parastoo Semnani"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/8xj9h27dbq3ge.bib",
  "cited_by": 0,
  "doi": "10.26434/chemrxiv.15008270/v1",
  "fulltext_url": null,
  "id": "8xj9h27dbq3ge",
  "issns": [],
  "landing_url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008270/v1",
  "language": null,
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "1703|Computational Drug Discovery Methods"
    ]
  ],
  "openalex": "W7208781236",
  "pdf_url": "https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008270/v1",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-03",
  "publisher": "American Chemical Society (ACS)",
  "ris": "https://research.draconitic.com/paper/8xj9h27dbq3ge.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:17Z",
      "native_id": "10.26434/chemrxiv.15008270",
      "source": "chemrxiv",
      "url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008270/v2"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "source:chemrxiv",
      "subfield": "general"
    }
  ],
  "title": "TAME: Element-wise Mixture-of-Experts Fusion for Data-Efficient and Interpretable Molecular Property Prediction",
  "title_text": "TAME: Element-wise Mixture-of-Experts Fusion for Data-Efficient and Interpretable Molecular Property Prediction",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:17Z",
  "url": "https://research.draconitic.com/paper/8xj9h27dbq3ge",
  "venue": "ChemRxiv",
  "versions": []
}