让微控制器跑CNN：压缩55倍，2KB内存也能部署
Deep Microcompression: Structured Pruning and Bit-packed Quantization for Microcontrollers
- 结构化剪枝+量化训练+位打包，实现55.8倍权重压缩
- 在LeNet-5上保持98.77%准确率，二进制大小缩小3倍
- 首次在2KB SRAM的ATmega328P上部署标准CNN，适合边缘设备
本文提出Deep Microcompression（DMC），一种面向裸金属微控制器的深度学习推理硬件感知流水线。DMC结合结构化剪枝、量化感知训练与固定长度位打包，实现对LeNet-5模型55.8×的权重压缩比（准确率达98.77%），生成无依赖的C语言库，具有确定性延迟。在RP2040（Cortex-M0+）上，相比TensorFlow Lite，DMC使二进制大小减少3倍且准确率相当。关键突破在于，DMC首次成功在仅2KB SRAM的ATmega328P上部署标准卷积神经网络，此前该设备被认为无法支持CNN推理。
原文摘要 · Abstract (English)
This paper introduces Deep Microcompression (DMC), a hardware-aware pipeline for deep learning inference on bare-metal microcontrollers. DMC integrates structured pruning, quantization-aware training, and fixed-length bit-packing to achieve a 55.8$\times$ weight compression ratio on LeNet-5 (98.77\% accuracy), generating a dependency-free C library with deterministic latency. On the RP2040 (Cortex-M0+), DMC reduces binary size by 3$\times$ versus TensorFlow Lite while matching its accuracy. Critically, DMC enables the first documented deployment of a standard CNN on the ATmega328P, a device constrained to 2KB SRAM, previously considered infeasible for CNN inference.
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。