Enhancing Generalization in Evolutionary Feature Construction for Symbolic Regression through Vicinal Jensen Gap Minimization
Hengzhe Zhang, Qi Chen, Bing Xue, Wolfgang Banzhaf, Mengjie Zhang · 2026
Read Enhancing Generalization in Evolutionary Feature Construction for Symbolic Regression through Vicinal Jensen Gap Minimization by Hengzhe Zhang, Qi Chen, Bing Xue, Wolfgang Banzhaf, Mengjie Zhang (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-2602-01510.txt.
Description
Hengzhe Zhang and co-authors improve generalization in evolutionary inputs for symbolic regression. Genetic programming builds inputs, yet overfitting limits use. They prove vicinal risk, gauged by noise, mixup, is bounded by empirical risk plus regularization: finite difference or the Jensen gap. Their framework optimizes empirical risk with the vicinal Jensen gap, curbing overfitting. As noise varies, they add a noise gauge tuning regularization. Against manifold intrusion, where added data creates unreal samples outside it, they check. 58 datasets show Jensen gap minimization beating others; tests with 15 methods show control winning.
Catalog record
| CREATOR | Hengzhe Zhang, Qi Chen, Bing Xue, Wolfgang Banzhaf, Mengjie Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 98675 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.01510