Random Attention Mechanism for Deep Quality Modeling and Online Prediction in Chemical Processes
Xie Ma, Lingjian Ye, Zhiqiang Ge · 2026
Read Random Attention Mechanism for Deep Quality Modeling and Online Prediction in Chemical Processes by Xie Ma, Lingjian Ye, Zhiqiang Ge (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-13261474.txt.
Description
Xie Ma, Lingjian Ye, and Zhiqiang Ge proposed in 2026 a random attention method for predicting product quality in chemical plants from process data. Quality measures are hard to sense directly, so software models infer them from easier readings. Deep latent variable models do this well, yet picking which hidden features to trust stayed murky.
Their method draws many random attention picks from the hidden features pooled across the deep model's layers. Each pick trains its own submodel for the same quality target. A fusion step then blends the submodels' answers into one prediction. On a debutanizer column, the method cut average prediction error by about a fifth against regular approaches. On a carbon dioxide absorption column, the cut reached about a third. The authors present the randomness as a feature, not noise, since varied views stabilize the estimate.
Catalog record
| CREATOR | Xie Ma, Lingjian Ye, Zhiqiang Ge |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |