A Semantic-Loss Function Modeling Framework With Task-Oriented Machine Learning Perspectives
Ti Ti Nguyen, Thanh-Dung Le, Vu Nguyen Ha, Hong-fu Chou, Geoffrey Eappen, Duc-Dung Tran, Hung Nguyen-Kha, Prabhu Thiruvasagam, Luis M. Garces-Socarras, Jorge L. Gonzalez-Rios, Juan C. Merlano-Duncan, Symeon Chatzinotas · 2025
Read A Semantic-Loss Function Modeling Framework With Task-Oriented Machine Learning Perspectives by Ti Ti Nguyen, Thanh-Dung Le, Vu Nguyen Ha, Hong-fu Chou, Geoffrey Eappen, Duc-Dung Tran, Hung Nguyen-Kha, Prabhu Thiruvasagam, Luis M. Garces-Socarras, Jorge L. Gonzalez-Rios, Juan C. Merlano-Duncan, Symeon Chatzinotas (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-2503-09903.txt.
Description
Ti Ti Nguyen and co-authors model meaning loss when Earth images are squeezed for satellite links, in work from twenty twenty-five. Satellites gather rich data but face tight bandwidth and delay limits. Sending raw data in full often proves impossible in real time. Meaning-first sending ships only what the task needs to decide. That shift needs a clear account of how meaning is lost.
Two loss types are split out and fitted from real data. Source loss tracks harm from processing the raw images themselves. Channel loss tracks harm from sending under real link limits. Four task models grade accuracy on lossy images at varied settings. Compression rates and channel states sweep a broad grid in the tests. The fits give a working frame for loss-aware satellite design. Planners can trade bandwidth against task accuracy with open eyes.
Catalog record
| CREATOR | Ti Ti Nguyen, Thanh-Dung Le, Vu Nguyen Ha, Hong-fu Chou, Geoffrey Eappen, Duc-Dung Tran, Hung Nguyen-Kha, Prabhu Thiruvasagam, Luis M. Garces-Socarras, Jorge L. Gonzalez-Rios, Juan C. Merlano-Duncan, Symeon Chatzinotas |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-09903