German Vega

Aleksandr Bredikhin, Philippe Lalanda, German Vega

Human Activity Recognition (HAR) is a core task in pervasive computing systems, where models must operate under strict computational constraints while remaining robust to heterogeneous and evolving deployment conditions. Recent advances based on Transformer architectures have significantly improved recognition performance, but typically rely on task-specific models trained from scratch, resulting in high training cost, large data requirements, and limited adaptability to domain shifts. In this paper, we propose a paradigm shift that reuses large pretrained language models (LLMs) as generic temporal backbones for sensor-based HAR, instead of designing domain-specific Transformers. To bridge the modality gap between inertial time series and language models, we introduce a structured convolutional projection that maps multivariate accelerometer and gyroscope signals into the latent space of the LLM. The pretrained backbone is kept frozen and adapted using parameter-efficient Low-Rank Adaptation (LoRA), drastically reducing the number of trainable parameters and the overall training cost. Through extensive experiments on standard HAR benchmarks, we show that this approach enables rapid convergence, strong data efficiency, and robust cross-dataset transfer, particularly in low-data and few-shot settings. At the same time, our results highlight the complementary roles of convolutional frontends and LLMs, where local invariances are handled at the signal level while long-range temporal dependencies are captured by the pretrained backbone. Overall, this work demonstrates that LLMs can serve as a practical, frugal, and scalable foundation for adaptive HAR systems, opening new directions for reusing foundation models beyond their original language domain.

Irina Seregina, Philippe Lalanda, German Vega

Human Activity Recognition is a foundational task in pervasive computing. While recent advances in self-supervised learning and transformer-based architectures have significantly improved HAR performance, adapting large pretrained models to new domains remains a practical challenge due to limited computational resources on target devices. This papers investigates parameter-efficient fine-tuning techniques, specifically Low-Rank Adaptation (LoRA) and Quantized LoRA, as scalable alternatives to full model fine-tuning for HAR. We propose an adaptation framework built upon a Masked Autoencoder backbone and evaluate its performance under a Leave-One-Dataset-Out validation protocol across five open HAR datasets. Our experiments demonstrate that both LoRA and QLoRA can match the recognition performance of full fine-tuning while significantly reducing the number of trainable parameters, memory usage, and training time. Further analyses reveal that LoRA maintains robust performance even under limited supervision and that the adapter rank provides a controllable trade-off between accuracy and efficiency. QLoRA extends these benefits by reducing the memory footprint of frozen weights through quantization, with minimal impact on classification quality.