KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition
The article discusses a study on Kolmogorov-Arnold Networks (KANs) and their application in improving human activity recognition (HAR) using inertial measurement units (IMUs). The authors propose a hybrid architecture that combines KANs with multi-layer perceptrons (MLPs) to enhance performance on real-world datasets. Their findings indicate that this approach significantly outperforms traditional models in terms of accuracy and robustness.
- ▪Kolmogorov-Arnold Networks (KANs) excel at learning complex functions on clean data but struggle with noisy datasets.
- ▪The proposed hybrid KAN-MLP model achieved a 5.33% improvement in average macro F1 score compared to pure MLP models.
- ▪Integrating the hybrid strategy into existing HAR architectures consistently enhances their performance.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.19031 (cs) [Submitted on 18 May 2026] Title:KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition Authors:Mengxi Liu, Sizhen Bian, Vitor Fortes, Francisco Calatrava Nicolas, Daniel Geißler, Maximilian Kiefer-Emmanouilidis, Bo Zhou, Paul Lukowicz View a PDF of the paper titled KAN-MLP-Mixer: A comprehensive investigation of the usage of Kolmogorov-Arnold Networks (KANs) for improving IMU-based Human Activity Recognition, by Mengxi Liu and 7 other authors View PDF HTML (experimental) Abstract:Kolmogorov-Arnold Networks (KANs) have demonstrated an exceptional ability to learn complex functions on clean, low-dimensional data but struggle to maintain…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
