@inproceedings{a53fbb1ee0d14c6c8bba9633531ee9d1,
title = "Local Reranking for Retrieval-based Heterogeneous Analog/RF Design",
abstract = "This paper studies retrieval-based inverse design for heterogeneous analog/RF systems on the AICircuit benchmark. Instead of predicting a unique parameter vector, the proposed method selects from a simulated design library the candidate whose achieved specifications best match a target query. We focus on the benchmark's heterogeneous Transmitter and Receiver, which capture stronger coupling than isolated circuit blocks. The proposed pipeline combines Euclidean top- K retrieval in normalized specification space with a learned local reranker. We also test a global triplet-loss retrieval model. Results show that local reranking is more effective for the harder Transmitter case, while Receiver achieves very low error under Euclidean retrieval.",
keywords = "Analog/RF design, Design retrieval, Heterogeneous systems, Inverse design",
author = "Ali Al-Zawqari and Ali Safa and Dries Peumans and Gerd Vandersteen",
note = "Publisher Copyright: {\textcopyright} 2026 IEEE.; 22nd International Conference on Synthesis, Modeling, Analysis and Simulation Methods, and Applications to Circuits Design, SMACD 2026 ; Conference date: 29-06-2026 Through 02-07-2026",
year = "2026",
doi = "10.1109/SMACD70206.2026.11647823",
language = "English",
series = "2026 22nd International Conference on Synthesis, Modeling, Analysis and Simulation Methods, and Applications to Circuits Design, SMACD 2026",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "2026 22nd International Conference on Synthesis, Modeling, Analysis and Simulation Methods, and Applications to Circuits Design, SMACD 2026",
address = "United States",
}