Search papers, labs, and topics across Lattice.
This paper addresses the problem of performance degradation in wearable human activity recognition (WHAR) models due to cross-user distribution shifts by proposing a novel test-time adaptation (TTA) framework called SIGHT. SIGHT leverages the temporal structure inherent in WHAR streams, using feature-conditioned inference to determine when to preserve temporal inertia and refine predictions during transitions. Experiments on real-world datasets demonstrate that SIGHT outperforms existing TTA baselines while reducing computational and memory costs, enabling real-time edge deployment.
Exploiting temporal continuity and feature deviations in wearable sensor data lets you adapt activity recognition models on the fly, boosting accuracy while slashing compute costs.
Wearable human activity recognition (WHAR) models often suffer from performance degradation under real-world cross-user distribution shifts. Test-time adaptation (TTA) mitigates this degradation by adapting models online using unlabeled test streams, yet existing methods largely inherit assumptions from vision tasks and underexploit the inherent inter-window temporal structure in WHAR streams. In this paper, we revisit such temporal structure as a feature-conditioned inference signal rather than merely an output-space smoothing prior. We derive the insight that temporal continuity and observation-induced feature deviations provide complementary cues for determining when to preserve or release temporal inertia and where to route prediction refinement during likely transitions. Building upon this insight, we propose SIGHT, a lightweight and backpropagation-free TTA framework for WHAR, enabling real-time edge deployment. SIGHT estimates predictive surprise by comparing the current feature with a prototype-based expected state, and then uses the resulting feature deviation to guide geometry-aware transition routing based on prototype alignment and stream-level marginal habit tracking. Evaluations on real-world datasets confirm that SIGHT outperforms existing TTA baselines while reducing computational and memory costs.
