arXiv subjects

Maximilian Burzer

Publications and source records attributed to Maximilian Burzer.

4 recordsLinked to original sources

arXiv2026

WHAR Arena: Benchmarking the State of the Art in Efficient Wearable Human Activity Recognition

Deep learning has become the dominant paradigm in Wearable Human Activity Recognition (WHAR), yet progress is obscured by a comparability crisis. Results are often reported using inconsistent datasets, custom data processing, and varying evaluation protocols, making state-of-the-art claims fragile. We address this with a large-scale, open-source benchmark that integrates 30 diverse datasets under standardized processing, unified model interfaces, and a shared cross-subject evaluation protocol. Evaluating 17 representative architectures across 4760 training runs, we jointly measure predictive performance alongside on-device latency, peak memory, and model size on an Android reference device. Our results reveal that the WHAR state of the art is distributed rather than dominated by a single architecture. While CNN-HAR achieves the highest mean macro-F1, top-performing models cluster tightly, indicating contemporary architectures have converged near a predictive performance ceiling. When accounting for deployment efficiency, compact neural models, such as TinierHAR, and classical Random Forests define the practically relevant Pareto frontier, whereas larger recurrent and hybrid models incur high hardware costs without corresponding performance gains. Consequently, while predictive performance has plateaued, substantial potential for future progress remains in optimizing deployment efficiency and improving adaptation to domain shifts. We release our full framework to support transparent reuse and extension.

cs.LG↗

arXiv2026

Uncertainty-Aware (Un)Supervised Few-Shot User Adaptation for On-Device Personalized Human Activity Recognition

Sensor-based Human Activity Recognition (HAR) models often degrade on unseen users due to domain shifts caused by individual movement patterns and sensor placement. Practical wearable HAR systems therefore require personalization methods that are lightweight, applicable whether calibration data is labeled, unlabeled, or unavailable, and robust under limited calibration. We present a gradient-free framework that repurposes pretrained HAR classifiers as Prototypical Networks using using prior prototypes, which preserve zero-shot performance and regularize adaptation. For labeled calibration, we introduce closed-form Bayesian prototype estimation and extend the same principle to unlabeled calibration. With only 3 seconds of calibration data per activity (one shot), supervised adaptation improves macro-F1 by +2.76 to +33.44 percentage points across four datasets, while unsupervised adaptation improves by +0.56 to +32.13 points. Since adaptation requires only closed-form prototype updates, the framework enables efficient and robust on-device personalization of preexisting HAR classifiers.

cs.LG↗

arXiv2026

CHOMP: Multimodal Chewing Side Detection with Earphones

Chewing side preference (CSP) has been identified both as a risk factor for temporomandibular disorders (TMD) and behavioral manifestation. Despite TMDs affecting roughly one third of the global population, assessment mainly relies on clinical examinations and self-reports, offering limited insight into everyday jaw function. Continuous CSP monitoring could provide an objective proxy for functional asymmetries. Prior wearable approaches, however, mostly use specialized form factors and demonstrate limited performance. We therefore present CHOMP, the first system for chewing side detection using earphones. Employing OpenEarable 2.0, we collected data from 20 participants with microphones, a bone-conduction microphone, IMU, PPG, and a pressure sensor across eleven foods, five non-chewing activities, and three noise conditions. We apply the Continuous Wavelet Transform to each sensing modality and use the resulting multi-channel scalograms as inputs to CNN-based classifiers. Microphones achieve the strongest single-sensor unit performance, with median F1 scores of 94.5% in leave-one-food-out (LOFO) and 92.6% in leave-one-subject-out (LOSO) cross-validations. Fusing sensing modalities further improves performance to 97.7% for LOFO and 95.4% for LOSO, with additional evaluations under noise interference indicating robust performance. Our results establish earphones as a practical platform for continuous CSP monitoring, enabling clinicians and patients to assess jaw function in everyday life.

cs.HC↗

arXiv2025

WHAR Datasets: An Open Source Library for Wearable Human Activity Recognition

The lack of standardization across Wearable Human Activity Recognition (WHAR) datasets limits reproducibility, comparability, and research efficiency. We introduce WHAR datasets, an open-source library designed to simplify WHAR data handling through a standardized data format and a configuration-driven design, enabling reproducible and computationally efficient workflows with minimal manual intervention. The library currently supports 9 widely-used datasets, integrates with PyTorch and TensorFlow, and is easily extensible to new datasets. To demonstrate its utility, we trained two state-of-the-art models, TinyHar and MLP-HAR, on the included datasets, approximately reproducing published results and validating the library's effectiveness for experimentation and benchmarking. Additionally, we evaluated preprocessing performance and observed speedups of up to 3.8x using multiprocessing. We hope this library contributes to more efficient, reproducible, and comparable WHAR research.

cs.HC↗