Role of Language-Derived Semantics in Knowledge Distillation on Wearable Sensor Data.
- 1Department of Computer Science and Engineering, Seoul National University of Science and Technology, Seoul 01811, Republic of Korea.
- 2Geometric Media Lab, The GAME School, Arizona State University, Tempe, AZ 85281, USA.
Abstract
Wearable sensor-based human activity recognition (HAR) has been widely studied for understanding human activities from time-series data. However, wearable sensor data exhibit substantial inter-individual variations, while the signal differences between activities can be relatively subtle. These characteristics make it challenging to learn effective representations for activity recognition. Considering these limitations, human cognition provides a useful motivation, as raw signals are interpreted together with prior knowledge and contextual expectations, suggesting that language-derived semantics can complement the information contained in time-series data. However, incorporating such semantic information into wearable systems requires careful consideration of the computational and memory constraints of edge devices. In this regard, knowledge distillation (KD) enables richer information to be exploited during training while retaining a compact student model for inference. Nevertheless, how language-derived semantics contribute to KD on wearable sensor data remains underexplored. In this study, we systematically investigate how textual semantics can be incorporated into KD through a Semantic Guidance Module (SGM), considering three different strategies: teacher-side (T-SG), student-side (S-SG), and both-side (B-SG). We examine these strategies through various evaluations, including analyses of teacher-student architectures, prompt types, and window lengths. Our findings demonstrate that language-derived semantics provide complementary knowledge for student learning and offer insights into how semantic information can be effectively incorporated into KD for wearable sensor-based HAR.
Related
- Knowledge distillation-based lightweight generative large language models for standardized quality control of Chinese radiology reports
- A Statistical Perspective on Knowledge Distillation: Foundations, Classical Methods, and Large Language Model Extensions
- Through the Looking Glass: Eye Tracking as a Cognitive Benchmark for Language-Model Semantics and Observer-Specific Semantic Alignment
- Lightweight anomaly detection via knowledge distillation from ensemble teachers
- Hybrid multi-model framework with knowledge distillation for precise and efficient sickle cell detection
BibTeX
@article{kang2026role,
title = {{Role of Language-Derived Semantics in Knowledge Distillation on Wearable Sensor Data.}},
author = {Min Chan Kang and Jae Chan Jeong and Pavan Turaga and Eun Som Jeon},
journal = {Biomimetics (Basel, Switzerland)},
year = {2026},
month = oct,
doi = {10.3390/biomimetics11100729},
url = {https://europepmc.org/article/MED/42930281},
note = {CC BY},
}
RIS
TY - JOUR
TI - Role of Language-Derived Semantics in Knowledge Distillation on Wearable Sensor Data.
AU - Min Chan Kang
AU - Jae Chan Jeong
AU - Pavan Turaga
AU - Eun Som Jeon
PY - 2026
DA - 2026/10/12
JO - Biomimetics (Basel, Switzerland)
DO - 10.3390/biomimetics11100729
UR - https://europepmc.org/article/MED/42930281
AB - Wearable sensor-based human activity recognition (HAR) has been widely studied for understanding human activities from time-series data. However, wearable sensor data exhibit substantial inter-individual variations, while the signal differences between activities can be relatively subtle. These characteristics make it challenging to learn effective representations for activity recognition. Considering these limitations, human cognition provides a useful motivation, as raw signals are interpreted together with prior knowledge and contextual expectations, suggesting that language-derived semantics can complement the information contained in time-series data. However, incorporating such semantic information into wearable systems requires careful consideration of the computational and memory constraints of edge devices. In this regard, knowledge distillation (KD) enables richer information to be exploited during training while retaining a compact student model for inference. Nevertheless, how language-derived semantics contribute to KD on wearable sensor data remains underexplored. In this study, we systematically investigate how textual semantics can be incorporated into KD through a Semantic Guidance Module (SGM), considering three different strategies: teacher-side (T-SG), student-side (S-SG), and both-side (B-SG). We examine these strategies through various evaluations, including analyses of teacher-student architectures, prompt types, and window lengths. Our findings demonstrate that language-derived semantics provide complementary knowledge for student learning and offer insights into how semantic information can be effectively incorporated into KD for wearable sensor-based HAR.
SN - 2313-7673
ER -