Clone-Robust Weights in Metric Spaces: Handling Redundancy Bias in Benchmark Aggregation
Damien Berriaud, Roger Wattenhofer · 2025
Read Clone-Robust Weights in Metric Spaces: Handling Redundancy Bias in Benchmark Aggregation by Damien Berriaud, Roger Wattenhofer (2025) 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-2502-03576.txt.
Description
Damien Berriaud and Roger Wattenhofer ask how to weight items so that near-duplicates cannot hijack the total. The problem shows up when tasks are pooled into a benchmark, when data points drive domain adaptation, and when opinions feed a voting advice tool. In each case, clones of one item can stack the deck. Their 2025 paper builds a theory for weights that resist this.
Clone-proof weighting makes similar items share importance instead of each claiming a full share. The framework grows out of the maximum uncertainty principle, extended to general metric spaces. Three axioms guide the design: symmetry, continuity, and clone-proofness. The authors then tackle when such weights exist. Benchmark designers get a principled defense against redundancy dressed as consensus.
Catalog record
| CREATOR | Damien Berriaud, Roger Wattenhofer |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 136 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-03576