Recommending Pre-Trained Models for IoT Devices

2024-12-25

AI Summary

Existing pre-trained model (PTM) selection methods (e.g., LogME, LEEP) for resource-constrained IoT devices ignore hardware constraints—such as memory, compute capacity, and power—leading to deployment failures. Method: This paper proposes a hardware-aware PTM recommendation framework that explicitly incorporates these constraints into the selection pipeline. It introduces (i) a lightweight hardware feature extraction module, (ii) a cross-platform joint prediction model for inference latency and memory footprint, and (iii) a task–hardware co-adaptability evaluation framework. Contribution/Results: Evaluated across eight mainstream IoT chip platforms, the method achieves a 27% improvement in recommendation accuracy and reduces end-to-end decision time by 90%, significantly enhancing PTM adaptability and deployment efficiency in edge-IoT scenarios.

Technology Category

• Machine Learning: Hardware-aware ML
• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

The availability of pre-trained models (PTMs) has enabled faster deployment of machine learning across applications by reducing the need for extensive training. Techniques like quantization and distillation have further expanded PTM applicability to resource-constrained IoT hardware. Given the many PTM options for any given task, engineers often find it too costly to evaluate each model's suitability. Approaches such as LogME, LEEP, and ModelSpider help streamline model selection by estimating task relevance without exhaustive tuning. However, these methods largely leave hardware constraints as future work-a significant limitation in IoT settings. In this paper, we identify the limitations of current model recommendation approaches regarding hardware constraints and introduce a novel, hardware-aware method for PTM selection. We also propose a research agenda to guide the development of effective, hardware-conscious model recommendation systems for IoT applications.

Problem

• Internet of Things
• Pre-trained Models
• Hardware Constraints

Innovation

• IoT devices
• hardware limitations
• model selection method