Learning ORDER-Aware Multimodal Representations for Composite Materials Design
Xinyao Li, Hangwei Qian, Jingjing Li, Ivor Tsang · 2026
Read Learning ORDER-Aware Multimodal Representations for Composite Materials Design by Xinyao Li, Hangwei Qian, Jingjing Li, Ivor Tsang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02513.txt.
Description
Xinyao Li and 4 co-authors design a learning system for composite materials, whose design spaces are continuous rather than graph-like, in a 2026 arXiv paper. Artificial intelligence excels at crystals and polymers, where discrete graphs capture structure. Composites break that pattern: fiber distributions matter, simple descriptors like fiber volume miss them, and data is scarce.
Their system, ORDER, aligns image and tabular data while treating ordinality as a core principle. Materials with similar target properties are forced to sit near each other in the learned representation space. That preserves the continuous nature of composite properties and allows interpolation between sparsely observed designs. Physics-based surrogate signals supply ordering information without full property labels. On nanofiber and carbon fiber data sets, ORDER beat alignment-oriented and property-aware rivals at prediction, cross-modal retrieval, and microstructure generation.
Catalog record
| CREATOR | Xinyao Li, Hangwei Qian, Jingjing Li, Ivor Tsang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 60 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-02513