EUVIMEDTutustu EUVIMEDiin
Open access · CC BY 4.0

Explainable Kolmogorov–Arnold Networks for Zero-Shot Human Activity Recognition on TinyML Edge Devices

Ismail Lamaakal, Chaymae Yahyati, Yassine Maleh, Khalid El Makkaoui, Ibrahim Ouahbi

Machine Learning and Knowledge Extraction · 2026-02-26 · 8 (3) · 55

DOI: 10.3390/make8030055 ↗

At a glance

Journal article in Machine Learning and Knowledge Extraction, published online on 2026-02-26. Written by Ismail Lamaakal and 4 co-authors. Cited 9 times so far (Crossref); 62 references. The full text is freely available (open access, CC BY 4.0).

TiivistelmäBibliografiset tiedotCiteRelated studies

Abstract (original language)

Human Activity Recognition (HAR) on wearable and IoT devices must jointly satisfy four requirements: high accuracy, the ability to recognize previously unseen activities, strict memory and latency constraints, and interpretable decisions. In this work, we address all four by introducing an explainable Kolmogorov–Arnold Network for Human Activity Recognition (TinyKAN-HAR) with a zero-shot learning (ZSL) module, designed specifically for TinyML edge devices. The proposed KAN replaces fixed activation functions by learnable one-dimensional spline operators applied after linear mixing, yielding compact yet expressive feature extractors whose internal nonlinearities can be directly visualized. On top of the KAN latent space, we learn a semantic projection and cosine-based compatibility function that align sensor features with class-level semantic embeddings, enabling both pure and generalized zero-shot recognition of unseen activities. We evaluate our method on three benchmark datasets (UCI HAR, WISDM, PAMAP2) under subject-disjoint and zero-shot splits. TinyKAN-HAR consistently achieves over 97% macro-F1 on seen classes and over 96% accuracy on unseen activities, with harmonic mean above 96% in the generalized ZSL setting, outperforming CNN, LSTM and Transformer-based ZSL baselines. For explainability, we combine gradient-based attributions, SHAP-style global relevance scores and inspection of the learned spline functions to provide sensor-level, temporal and neuron-level insights into each prediction. After 8-bit quantization and TinyML-oriented optimizations, the deployed model occupies only 145 kB of flash and 26 kB of RAM, and achieves an average inference latency of 4.1 ms (about 0.32 mJ per window) on a Cortex-M4F-class microcontroller, while preserving accuracy within 0.2% of the full-precision model. These results demonstrate that explainable, zero-shot HAR with near state-of-the-art accuracy is feasible on severely resource-constrained TinyML edge devices.

Cite

APA

Lamaakal, I., Yahyati, C., Maleh, Y., Makkaoui, K. E., & Ouahbi, I. (2026). Explainable Kolmogorov–Arnold Networks for Zero-Shot Human Activity Recognition on TinyML Edge Devices. Machine Learning and Knowledge Extraction, 8(3), 55. https://doi.org/10.3390/make8030055

Vancouver

Lamaakal I, Yahyati C, Maleh Y, Makkaoui KE, Ouahbi I. Explainable Kolmogorov–Arnold Networks for Zero-Shot Human Activity Recognition on TinyML Edge Devices. Machine Learning and Knowledge Extraction. 2026;8(3):55. doi:10.3390/make8030055

BibTeX
@article{Lamaakal2026,
  author = {Lamaakal, Ismail and Yahyati, Chaymae and Maleh, Yassine and Makkaoui, Khalid El and Ouahbi, Ibrahim},
  title = {Explainable Kolmogorov–Arnold Networks for Zero-Shot Human Activity Recognition on TinyML Edge Devices},
  journal = {Machine Learning and Knowledge Extraction},
  year = {2026},
  volume = {8},
  number = {3},
  pages = {55},
  doi = {10.3390/make8030055}
}
Start for free →