Reproduce “ECC-SNN: Cost-Effective Edge-Cloud Collaboration for Spiking Neural Networks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Di Yu, Changze Lv, Xin Du, Linshan Jiang, Wentao Tong, Zhenyu Liao, Xiaoqing Zheng, Shuiguang Deng
A framework for cost-effective edge-cloud collaboration, ECC-SNN, integrates cloud-based Artificial Neural Networks (ANNs) with energy-efficient edge Spiking Neural Networks (SNNs) to enhance performance and adaptability in resource-constrained environments. Developed by researchers from Zhejiang University, Fudan University, and National University of Singapore, it achieves an average 79.4% energy consumption reduction and 39.1% processing latency reduction by selectively offloading ambiguous tasks to the cloud and enabling on-device incremental learning for SNNs.
Key insights
- Integrating high-accuracy cloud ANNs with energy-efficient edge SNNs through knowledge distillation effectively bridges the performance gap of SNNs.
- Intelligent collaborative inference, which selectively offloads ambiguous samples to the cloud, significantly reduces energy consumption and latency.
- On-device, exemplar-free incremental learning for SNNs allows edge models to continuously adapt to dynamic data distributions while mitigating catastrophic forgetting, enhancing autonomy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.