用少量样本和极小参数量，修复阻变存内计算的精度退化问题。
Efficient Calibration for RRAM-based In-Memory Computing using DoRA
- 基于DoRA的校准框架，仅更新少量参数补偿权重影响。
- 仅用10个样本即可恢复69.53%准确率，仅更新2.34%参数。
- 避免现场写入RRAM，实现高效、快速、可靠的校准。
阻变存内计算（RIMC）为边缘AI提供超高效计算，但受RRAM电导漂移影响导致精度下降。传统重训练方法受限于RRAM高功耗、写延迟和耐久性问题。本文提出基于DoRA的校准框架，通过在SRAM中存储极少校准参数，补偿关键权重影响，无需修改RRAM权重。该方法避免现场写入RRAM，实现节能、快速、可靠的校准。在基于RIMC的ResNet50（ImageNet-1K）上实验表明，仅使用10个校准样本即可恢复69.53%准确率，且仅更新2.34%的参数。
原文摘要 · Abstract (English)
Resistive In-Memory Computing (RIMC) offers ultra-efficient computation for edge AI but faces accuracy degradation due to RRAM conductance drift over time. Traditional retraining methods are limited by RRAM's high energy consumption, write latency, and endurance constraints. We propose a DoRA-based calibration framework that restores accuracy by compensating influential weights with minimal calibration parameters stored in SRAM, leaving RRAM weights untouched. This eliminates in-field RRAM writes, ensuring energy-efficient, fast, and reliable calibration. Experiments on RIMC-based ResNet50 (ImageNet-1K) demonstrate 69.53% accuracy restoration using just 10 calibration samples while updating only 2.34% of parameters.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。