상세 보기
Weight Sharing for Array-Efficient CNN Inference in Compute-in-Memory Architectures
- Rhe, Johnny;
- Ko, Jong Hwan
Citations
SCOPUS
0초록
In this work, we propose a weight sharing method for compute-in-memory (CIM)-based convolutional neural network (CNN) inference, which reduces the number of mapped weights without relying on pruning. By sharing weights and layer-wise scale factors, our approach reduces overall CIM array usage while maintaining model accuracy. Evaluations on ResNet-20 with 32×32 arrays demonstrate that our method achieves a 28% reduction in array usage with less than 1% accuracy loss.
키워드
compute-in-memory (CIM); weight mapping
- 제목
- Weight Sharing for Array-Efficient CNN Inference in Compute-in-Memory Architectures
- 저자
- Rhe, Johnny; Ko, Jong Hwan
- 발행일
- 2025
- 유형
- Conference Paper
- 저널명
- International SoC Design Conference 2025, ISOCC 2025 - Proceedings of Technical Papers