Inverse Design in Distributed Circuits Using Single-Step Reinforcement Learning
Jiayu Li, Masood Mortazavi, Ning Yan, Yihong Ma, Reza Zafarani · 2025
Read Inverse Design in Distributed Circuits Using Single-Step Reinforcement Learning by Jiayu Li, Masood Mortazavi, Ning Yan, Yihong Ma, Reza Zafarani (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-2506-08029.txt.
Description
Jiayu Li and co-workers present DCIDA, a learning framework that designs distributed circuits to meet a target response, in a 2025 paper. Inverse design must search layouts that vary in shape and placement, and real checks of a design may not be differentiable. Older methods lean on fixed grids or templates.
DCIDA decides every design factor in one compound action, sampling from jointly trained probability distributions. A special one-to-one map turns raw samples into unique physical layouts, so the system learns how choices depend on each other. Its transformer-based policy cut design error well below prior methods. The advantage grew on harder target responses. The authors frame it as practical design automation.
Catalog record
| CREATOR | Jiayu Li, Masood Mortazavi, Ning Yan, Yihong Ma, Reza Zafarani |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-08029