Special Research Session: SatReg: Regression-based Neural Architecture Search for Lightweight Satellite Image Segmentation Conference: 63rd ACM/IEEE Design Automation Conference; July 26–29, 2026; Long Beach, CA, USA 63rd ACM/IEEE Design Automation Conference (DAC ’26), July 26–29, 2026, Long Beach, CA, USA DOI: 10.1145/3770743.3807769 ISBN: 979-8-4007-2254-7/2026/07 CCS: Hardware Power and energy CCS: Computing methodologies Machine learning
Edward Humes, Tinoosh Mohsenin · 2026
Read Special Research Session: SatReg: Regression-based Neural Architecture Search for Lightweight Satellite Image Segmentation Conference: 63rd ACM/IEEE Design Automation Conference; July 26–29, 2026; Long Beach, CA, USA 63rd ACM/IEEE Design Automation Conference (DAC ’26), July 26–29, 2026, Long Beach, CA, USA DOI: 10.1145/3770743.3807769 ISBN: 979-8-4007-2254-7/2026/07 CCS: Hardware Power and energy CCS: Computing methodologies Machine learning by Edward Humes, Tinoosh Mohsenin (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-2604-10306.txt.
Description
Humes and Mohsenin present SatReg, a tuning framework for lightweight remote-sensing segmentation on edge platforms. Using a teacher, they shrink the search to two width variables, profile student models on a Jetson Orin Nano, and fit surrogate models for accuracy, latency, and power. Distillation trains the sampled students efficiently. The surrogates pick best settings for deployment targets without exhaustive search. The chosen variables affect accuracy and hardware cost differently, making reduced regression practical for hybrid CNN-Mamba segmentation in space systems.
Catalog record
| CREATOR | Edward Humes, Tinoosh Mohsenin |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15350 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.10306