PRIOT: Pruning-Based Integer-Only Transfer Learning for Embedded Systems

2025-03-21

IEEE Embedded Systems Letters

AI Summary

Traditional fine-tuning fails under static quantization on resource-constrained embedded devices lacking floating-point units (e.g., Raspberry Pi Pico), while dynamic scale computation incurs prohibitive overhead. Method: We propose the first integer-only edge-side transfer learning framework, abandoning weight updates in favor of a novel integer training paradigm based on edge pruning. We introduce Edge-Popup, a heuristic pruning mechanism, and its memory-optimized variant PRIOT-S—enabling efficient, low-memory integer fine-tuning under static quantization. Contributions/Results: On rotated MNIST and rotated CIFAR-10, our method improves accuracy by 8.08–33.75 percentage points. PRIOT-S reduces memory footprint significantly, with <1% accuracy degradation, and supports C/C++ deployment on ARM Cortex-M0+ microcontrollers. This work establishes the first practical, fully integer, statically quantized fine-tuning approach for ultra-low-power edge devices.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Learning & Optimization for CV
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Security and Privacy: Privacy-enhancing technologies
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

On-device transfer learning is crucial for adapting a common backbone model to the unique environment of each edge device. Tiny microcontrollers, such as the Raspberry Pi Pico, are key targets for on-device learning but often lack floating-point units, necessitating integer-only training. Dynamic computation of quantization scale factors, which is adopted in former studies, incurs high computational costs. Therefore, this study focuses on integer-only training with static scale factors, which is challenging with existing training methods. We propose a new training method named PRIOT, which optimizes the network by pruning selected edges rather than updating weights, allowing effective training with static scale factors. The pruning pattern is determined by the edge-popup algorithm, which trains a parameter named score assigned to each edge instead of the original parameters and prunes the edges with low scores before inference. Additionally, we introduce a memory-efficient variant, PRIOT-S, which only assigns scores to a small fraction of edges. We implement PRIOT and PRIOT-S on the Raspberry Pi Pico and evaluate their accuracy and computational costs using a tiny CNN model on the rotated MNIST dataset and the VGG11 model on the rotated CIFAR-10 dataset. Our results demonstrate that PRIOT improves accuracy by 8.08 to 33.75 percentage points over existing methods, while PRIOT-S reduces memory footprint with minimal accuracy loss.

Problem

• Enables integer-only training on microcontrollers without floating-point units
• Reduces computational costs by using static quantization scale factors
• Improves accuracy and memory efficiency in on-device transfer learning

Innovation

• Pruning edges instead of updating weights
• Static scale factors for integer-only training
• Memory-efficient variant with selective scoring