PulseAugur
实时 21:21:57
English(EN) From Compression to Deployment: Real-Time and Energy-Efficient FastGRNN on Ultra-Constrained Microcontrollers

FastGRNN模型部署在微控制器上实现实时推理

研究人员开发了一个端到端的流程,将FastGRNN模型部署在超约束微控制器上,特别是Arduino (ATmega328P) 和 TI MSP430。该方法侧重于为小型、普遍存在的设备重构AI算法,与扩展模型规模的趋势形成对比。部署的模型仅占用566字节的权重,在HAPT测试集上实现了0.918的宏F1分数,并能维持实时50 Hz流式推理。 AI

影响 在低功耗、资源受限的边缘设备上实现实时AI推理,将AI的应用范围扩展到传统硬件之外。

排序理由 该集群包含一篇学术论文,详细介绍了在微控制器上部署AI模型的新方法。

在 arXiv cs.NE (Neural & Evolutionary) 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

FastGRNN模型部署在微控制器上实现实时推理

报道来源 [2]

  1. arXiv cs.LG TIER_1 English(EN) · Emre Can Kizilates ·

    From Compression to Deployment: Real-Time and Energy-Efficient FastGRNN on Ultra-Constrained Microcontrollers

    arXiv:2606.17249v1 Announce Type: cross Abstract: The dominant trajectory of modern machine learning has been to scale up: larger models, larger accelerators, larger memory budgets. Yet a multi-year global semiconductor supply constraint and the growing energy and carbon cost of …

  2. arXiv cs.NE (Neural & Evolutionary) TIER_1 English(EN) · Emre Can Kizilates ·

    From Compression to Deployment: Real-Time and Energy-Efficient FastGRNN on Ultra-Constrained Microcontrollers

    The dominant trajectory of modern machine learning has been to scale up: larger models, larger accelerators, larger memory budgets. Yet a multi-year global semiconductor supply constraint and the growing energy and carbon cost of always-online inference expose the fragility of th…