© 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to s
Ahmad Ismail, Raphaël Le Bidan, Elsa Dupraz, Charbel Abdel Nour · 2025
Read © 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to s by Ahmad Ismail, Raphaël Le Bidan, Elsa Dupraz, Charbel Abdel Nour (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-2502-10183.txt.
Description
Neural decoders can fix errors in transmitted data, but their training sets are usually brewed on the fly, endlessly and wastefully. Ahmad Ismail and co-authors argue that a fixed, carefully chosen dataset trains better decoders with fewer examples. Their study targets syndrome-based decoders, which work from error fingerprints rather than raw signals.
The authors spell out why fixed data beats dynamic generation for this task. The harder choice is which training targets to include. They propose several selection heuristics and test them head to head. With curated data, the decoders reached stronger performance while eating fewer examples. Code to reproduce every result is public. The lesson reaches beyond decoding: in some tasks, data curation is the cheapest performance upgrade available.
Catalog record
| CREATOR | Ahmad Ismail, Raphaël Le Bidan, Elsa Dupraz, Charbel Abdel Nour |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-10183