On-device edge learning for IoT data streams: a survey

2025-02-25

AI Summary

Continuous classification of IoT tabular data streams on resource-constrained edge devices demands online learning for neural networks and decision trees in open-world settings, confronting challenges including catastrophic forgetting, concept drift, and low data efficiency. Method: We propose (i) the first multi-dimensional continual learning evaluation framework tailored for TinyML, jointly assessing output performance and internal representation quality; (ii) a dynamic pruning–meta-learning co-adaptive mechanism for decision trees that explicitly balances stability and plasticity; and (iii) a cloud–edge–end data architecture constraint mapping model. Results: Experiments demonstrate that our approach significantly mitigates forgetting and enhances robustness to concept drift on edge devices. It establishes a reusable design paradigm and benchmarking toolkit for autonomous, online edge intelligence systems—advancing both algorithmic adaptability and system-level deployment feasibility under stringent hardware constraints.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Multiagent Systems: Multiagent Learning
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

This literature review explores continual learning methods for on-device training in the context of neural networks (NNs) and decision trees (DTs) for classification tasks on smart environments. We highlight key constraints, such as data architecture (batch vs. stream) and network capacity (cloud vs. edge), which impact TinyML algorithm design, due to the uncontrolled natural arrival of data streams. The survey details the challenges of deploying deep learners on resource-constrained edge devices, including catastrophic forgetting, data inefficiency, and the difficulty of handling IoT tabular data in open-world settings. While decision trees are more memory-efficient for on-device training, they are limited in expressiveness, requiring dynamic adaptations, like pruning and meta-learning, to handle complex patterns and concept drifts. We emphasize the importance of multi-criteria performance evaluation tailored to edge applications, which assess both output-based and internal representation metrics. The key challenge lies in integrating these building blocks into autonomous online systems, taking into account stability-plasticity trade-offs, forward-backward transfer, and model convergence.

Problem

• Explores continual learning methods for on-device training
• Highlights constraints impacting TinyML algorithm design
• Details challenges of deploying deep learners on edge devices

Innovation

• on-device edge learning
• continual learning methods
• multi-criteria performance evaluation