ADLGen: Synthesizing Symbolic, Event-Triggered Sensor Sequences for Human Activity Modeling

2025-05-23

AI Summary

Real-world daily activity recognition faces challenges including scarcity of real sensor data, privacy sensitivity, and high annotation costs. To address these, this paper proposes a physics-constrained symbolic sensor sequence generation method. The approach innovatively integrates symbolic temporal encoding with layout-aware sampling and establishes an LLM-driven “generate–evaluate–refine” closed-loop framework, ensuring full automation of logical, behavioral, and temporal consistency without manual intervention or environment-specific tuning. Built upon a decoder-only Transformer architecture, the method jointly models symbolic time encodings and spatial contextual dependencies. Experiments demonstrate that the generated sequences surpass baselines in statistical fidelity, semantic richness, and downstream activity recognition accuracy—achieving up to 12.7% improvement in F1-score over state-of-the-art synthetic data methods. Moreover, the synthesized data exhibits enhanced practical utility and strong cross-scenario generalization capability.

Technology Category

• Planning, Routing, and Scheduling: Activity and Plan Recognition
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Video Understanding & Activity Analysis

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Real world collection of Activities of Daily Living data is challenging due to privacy concerns, costly deployment and labeling, and the inherent sparsity and imbalance of human behavior. We present ADLGen, a generative framework specifically designed to synthesize realistic, event triggered, and symbolic sensor sequences for ambient assistive environments. ADLGen integrates a decoder only Transformer with sign based symbolic temporal encoding, and a context and layout aware sampling mechanism to guide generation toward semantically rich and physically plausible sensor event sequences. To enhance semantic fidelity and correct structural inconsistencies, we further incorporate a large language model into an automatic generate evaluate refine loop, which verifies logical, behavioral, and temporal coherence and generates correction rules without manual intervention or environment specific tuning. Through comprehensive experiments with novel evaluation metrics, ADLGen is shown to outperform baseline generators in statistical fidelity, semantic richness, and downstream activity recognition, offering a scalable and privacy-preserving solution for ADL data synthesis.

Problem

• Challenges in real-world ADL data collection due to privacy and cost
• Generating realistic sensor sequences for assistive environments
• Ensuring semantic and temporal coherence in synthetic ADL data

Innovation

• Decoder-only Transformer with symbolic temporal encoding
• Context and layout aware sampling mechanism
• LLM-based automatic generate-evaluate-refine loop