MUDAS: Mote-scale Unsupervised Domain Adaptation in Multi-label Sound Classification

2025-06-12

AI Summary

Unsupervised domain adaptation (UDA) for cross-domain multi-label sound classification on resource-constrained IoT devices remains challenging, as existing UDA methods are designed for single-label tasks and incur prohibitive computational overhead. Method: We propose the first lightweight unsupervised domain adaptation framework tailored for edge deployment. Its core innovations include: (1) a class-specific adaptive-threshold pseudo-labeling mechanism for dynamic selection and retraining of high-confidence samples; (2) multi-label diversity regularization to mitigate label coupling and noise propagation; and (3) end-to-end edge-aware model compression and deployment optimization. Results: Evaluated on the SONYC-UST dataset, our method significantly outperforms state-of-the-art UDA approaches in multi-label accuracy and robustness. It is the first to enable real-time, robust multi-label acoustic sensing on micro-scale IoT devices—achieving both low latency and high fidelity under strict memory and compute constraints.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• 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

Unsupervised Domain Adaptation (UDA) is essential for adapting machine learning models to new, unlabeled environments where data distribution shifts can degrade performance. Existing UDA algorithms are designed for single-label tasks and rely on significant computational resources, limiting their use in multi-label scenarios and in resource-constrained IoT devices. Overcoming these limitations is particularly challenging in contexts such as urban sound classification, where overlapping sounds and varying acoustics require robust, adaptive multi-label capabilities on low-power, on-device systems. To address these limitations, we introduce Mote-scale Unsupervised Domain Adaptation for Sounds (MUDAS), a UDA framework developed for multi-label sound classification in resource-constrained IoT settings. MUDAS efficiently adapts models by selectively retraining the classifier in situ using high-confidence data, minimizing computational and memory requirements to suit on-device deployment. Additionally, MUDAS incorporates class-specific adaptive thresholds to generate reliable pseudo-labels and applies diversity regularization to improve multi-label classification accuracy. In evaluations on the SONYC Urban Sound Tagging (SONYC-UST) dataset recorded at various New York City locations, MUDAS demonstrates notable improvements in classification accuracy over existing UDA algorithms, achieving good performance in a resource-constrained IoT setting.

Problem

• Adapting models to unlabeled environments with data shifts
• Enabling multi-label UDA for resource-constrained IoT devices
• Improving urban sound classification accuracy on low-power systems

Innovation

• Selective retraining with high-confidence data
• Class-specific adaptive thresholds for pseudo-labels
• Diversity regularization for multi-label accuracy