PERTURB-c: Correlation Aware Perturbation Explainability for Regression Techniques to Understand Retrieval Black-boxes
Jools D. Clarke, 1 Gordon Yip 2 and Nikolaos Nikolaou 1 1 Department of Physics and Astronomy, University College London, Gower Street, WC1E 6BT London, United Kingdom 2 NMES Faculty, King’s College London, Strand Building, Strand, WC2R 2LS London, United Kingdom · 2026
Read PERTURB-c: Correlation Aware Perturbation Explainability for Regression Techniques to Understand Retrieval Black-boxes by Jools D. Clarke, 1 Gordon Yip 2 and Nikolaos Nikolaou 1 1 Department of Physics and Astronomy, University College London, Gower Street, WC1E 6BT London, United Kingdom 2 NMES Faculty, King’s College London, Strand Building, Strand, WC2R 2LS London, United Kingdom (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-2601-21685.txt.
Description
Jools D. Clarke, Gordon Yip, and Nikolaos Nikolaou explain black-box regression models whose inputs have internal structure. Their framework perturbs the input and watches the output move. Unlike standard tools, it accounts for correlations inside the structured input.
Ignoring those correlations can credit the wrong part of the input. The team demonstrates the method on machine learning retrievals of an exoplanet's atmosphere from transit data. The planet used is WASP-107b, a well-studied case. Trustworthy explanations matter in astronomy, where a model's answer about a distant world cannot be checked by hand.
Catalog record
| CREATOR | Jools D. Clarke, 1 Gordon Yip 2 and Nikolaos Nikolaou 1 1 Department of Physics and Astronomy, University College London, Gower Street, WC1E 6BT London, United Kingdom 2 NMES Faculty, King’s College London, Strand Building, Strand, WC2R 2LS London, United Kingdom |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 79 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-21685