analiza aiqbench
KOALA advances human motion prediction by leveraging WiFi Channel State Information within a Koopman operator framework to linearize nonlinear dynamics and mitigate error accumulation inherent in autoregressive methods. This approach enables robust multi-horizon forecasting without relying on camera inputs, offering a significant step forward for privacy-preserving activity recognition systems. The method's ability to enforce dynamical consistency through novel loss functions represents a meaningful improvement over existing vision-based or naive regression baselines.
KOALA: Koopman Operator Learning for WiFi-Based Anticipatory Hum
arXiv:2608.15815v1 Announce Type: new Abstract: WiFi Channel State Information (CSI) has emerged as a privacy-preserving alternative to cameras for human pose estimation. However, existing approaches treat pose inference as an instantaneous regression problem and do not model temporal dynamics, making future motion prediction infeasible. Naively applying vision-based prediction methods compounds the estimation noise already present in CSI-derived poses, as autoregressive rollouts amplify errors at every step. We propose KOALA, the framework for human motion prediction directly from WiFi CSI, by lifting noisy CSI-derived pose sequences into a learned Koopman latent space where nonlinear dynamics become linear, enabling multi-horizon prediction via simple matrix-vector products without autoregressive iteration or error accumulation. A residual CSI-conditioned operator resolves the identity attractor problem inherent from Koopman formulations, and an anchor-delta prediction head eliminates the degenerate shortcut of copying the current pose across all horizons. To regularise the lifting and operator jointly, we introduce a Koopman Anchored Latent (KAL) loss that operates in the temporal-encoder feature space, enforcing dynamical consistency across prediction horizons without requiring contrastive, spectral, or auxiliary losses. Experiments on MM-Fi and WiPose show that KOALA achieves robust, consistent performance across both short- and long-term prediction horizons, outperforming all baselines by a substantial margin.
Abstract:WiFi Channel State Information (CSI) has emerged as a privacy-preserving alternative to cameras for human pose estimation. However, existing approaches treat pose inference as an instantaneous regression problem and do not model temporal dynamics, making future motion prediction infeasible. Naively applying vision-based prediction methods compounds the estimation noise already present in CSI-derived poses, as autoregressive rollouts amplify errors at every step. We propose KOALA, the framework for human motion prediction directly from WiFi CSI, by lifting noisy CSI-derived pose sequences into a learned Koopman latent space where nonlinear dynamics become linear, enabling multi-horizon prediction via simple matrix-vector products without autoregressive iteration or error accumulation. A residual CSI-conditioned operator resolves the identity attractor problem inherent from Koopman formulations, and an anchor-delta prediction head eliminates the degenerate shortcut of copying the current pose across all horizons. To regularise the lifting and operator jointly, we introduce a Koopman Anchored Latent (KAL) loss that operates in the temporal-encoder feature space, enforcing dynamical consistency across prediction horizons without requiring contrastive, spectral, or auxiliary losses. Experiments on MM-Fi and WiPose show that KOALA achieves robust, consistent performance across both short- and long-term prediction horizons, outperforming all baselines by a substantial margin.
Machine Learning (cs.LG)
arXiv:2608.15815 [cs.LG]
(or arXiv:2608.15815v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.15815 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 289 cuvinte
Subiecte conexe
Koopman Theory and Metaethics
LessWrong · 19 sept. 2026
SACK: Safe Active Continual Koopman Learning for Uncertain Systems with Contractive Guarantees
arXiv · 6 aug. 2026
Bilevel optimization for data-driven learning of Koopman embeddings using kernel-based autoencoders
arXiv · 9 oct. 2026
Mobile Multi-Robot Navigation under Runtime Uncertainty via Koopman Operator Learning and Nonlinear Model Predictive Control
arXiv · 15 sept. 2026