Representation Retrieval Learning for Heterogeneous Data Integration
Qi Xu, Annie Qu · 2026
Read Representation Retrieval Learning for Heterogeneous Data Integration by Qi Xu, Annie Qu (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/pmc-13645369.txt.
Description
Qi Xu and Annie Qu propose, in a 2026 paper, a framework for learning from data sets that refuse to line up. Modern collections come from many sources and senses at once, and they clash: measurements shift between sources, outcomes drift, and whole blocks go missing. Standard supervised learning sags under that mix. Their framework, called R2, keeps a dictionary of representation modules and lets each source pick its own sparse learner.
A new score, integrativeness, rates how well each module serves across sources. A penalty they design then favors the more integrative modules, and theory shows the framework's error bound tightens as that quality rises. Simulations back the design, and two real-world data sets confirm the gain in practice. The authors aim the method at big, messy, multi-source problems in science and industry alike.
Catalog record
| CREATOR | Qi Xu, Annie Qu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |