BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260402T024508Z
LOCATION:3006\, Level 3
DTSTART;TZID=America/Los_Angeles:20250623T103000
DTEND;TZID=America/Los_Angeles:20250623T120000
UID:dac_DAC 2025_sess139@linklings.com
SUMMARY:LLM/DL Driven Analog Circuit Design and Analysis
DESCRIPTION:Modern analog IC design faces challenges like increasing compl
 exity, optimization, PVT variations, and simulation bottlenecks. Tradition
 al methods struggle with large design spaces and reliability. LLMs and dee
 p learning technologies address these issues by automating tasks, managing
  PVT variations, and reducing simulation costs. This session will cover th
 e topics reinforcement learning for variation-aware designs, transfer lear
 ning for system-level optimization, universal neural simulators, GPU-accel
 erated passivity enforcement, and image-graph fusion for IR drop predictio
 n, demonstrating AI’s role in optimizing and enhancing analog IC design.\n
 \nGLOVA: Global and Local Variation-Aware Analog Circuit Design with Risk-
 Sensitive Reinforcement Learning\n\nAnalog/mixed-signal circuit design enc
 ounters significant challenges due to performance degradation from process
 , voltage, and temperature (PVT) variations. To achieve commercial-grade r
 eliability, iterative manual design revisions and extensive statistical si
 mulations are required. While several st...\n\n\nDongjun Kim, Junwoo Park,
  Chaehyeon Shin, and Jaeheon Jung (Korea University); Kyungho Shin, Seungh
 eon Baek, Sanghyuk Heo, Woongrae Kim, Inchul Jeong, and Joohwan Cho (SK hy
 nix); and Jongsun Park (Korea University)\n---------------------\nINSIGHT:
  A Universal Neural Simulator Framework for Analog Circuits with Autoregre
 ssive Transformers\n\nThis paper introduces INSIGHT, a data-efficient, ada
 ptive, high-fidelity, technology-agnostic universal neural simulator frame
 work that formulates analog performance prediction as an autoregressive se
 quence generation task to accurately predict performance across diverse ci
 rcuits. INSIGHT achieves t...\n\n\nSouradip Poddar (The University of Texa
 s at Austin), Youngmin Oh (Samsung), Yao Lai and Hanqing Zhu (The Universi
 ty of Texas at Austin), Bosun Hwang (Samsung), and David Z. Pan (The Unive
 rsity of Texas at Austin)\n---------------------\nG-SpNN: GPU-Accelerated 
 Passivity Enforcement for S-Parameter Modeling with Neural Networks\n\nThe
  increasing complexity of high-frequency circuits calls for efficient and 
 accurate passive macromodeling techniques.\nExisting passivity enforcement
  methods, including those in commercial tools, often encounter convergence
  issues or compromise accuracy.\nThe Domain-Alternated Optimization (DAO) 
 fra...\n\n\nLijie Zeng and Jiatai Sun (China University of Petroleum, Beij
 ing); Xiao Wu (Huada Empyrean Software Co. Ltd); Dan Niu (Southeast Univer
 sity); Tianshi Wang (University of California, Berkeley); Yibo Lin (Peking
  University); Zuochang Ye (Tsinghua University); and Zhou Jin (China Unive
 rsity of Petroleum, Beijing)\n---------------------\nA Novel Image-Graph H
 eterogeneous Fusion Framework for Static IR Drop Prediction\n\nIR drop ana
 lysis is crucial for ensuring the reliability and performance of integrate
 d circuits (ICs) but poses computational challenges as the IC designs grow
  larger, especially for ultra deep-submicron VLSI designs. Deep learnings 
 (DL) as the efficiency-promising solutions, mainly employ various C...\n\n
 \nDan Niu, Dekang Zhang, and Yichao Cao (Southeast University); Zhou Jin (
 China University of Petroleum, Beijing); Chao Wang and Yichao Dong (Southe
 ast University); and Changyin Sun (Anhui University)\n--------------------
 -\nGraph-Guided Transfer Learning to Boost the Efficiency of System-Level 
 Optimization of Analog/Mixed-Signal Circuits\n\nThis paper introduces a no
 vel graph-guided transfer learning approach to boost the efficiency of sys
 tem-level optimization of analog/mixed-signal circuits. The system-level o
 ptimization is based on Reinforcement Learning (RL) in combination with Gr
 aph Attention Networks (GAT). The results surpass st...\n\n\nJiaqi Wang an
 d Georges Gielen (KU Leuven)\n\nTopics: EDA\n\nTracks: EDA6: Analog CAD, S
 imulation, Verification and Test\n\nSession Chairs: Arindam Basu (School o
 f Computer Science and Engineering, Nanyang Technological University) and 
 Markus Olbrich (Leibniz University Hannover)
END:VEVENT
END:VCALENDAR