When Adaptation Hurts: Split Sensitivity and Person-Level Negative Transfer in Federated Wearable Onboarding
cs.LG, cs.DC
Submitted: 2026-08-18
Updated: 2026-08-18
Comments: 14 pages, 2 figures, 4 tables; supplementary material included. Code and supplementary scripts: https://github.com/TitanRahil/When-Adaptation-Hurts
Code: https://github.com/TitanRahil/When-Adaptation-Hurts
License: http://creativecommons.org/licenses/by-nc-sa/4.0/
The gist: Federated wearable models eventually serve people absent from source training, but favorable average accuracy does not establish that unlabeled onboarding helps each person.
Terminology
Abstract
Federated wearable models eventually serve people absent from source training, but favorable average accuracy does not establish that unlabeled onboarding helps each person. We evaluate six core onboarding strategies on five wearable datasets under a leakage-controlled protocol that fixes source checkpoints, estimates normalization from source data only, separates calibration from evaluation recordings, and performs inference over held-out people rather than windows, devices, or random seeds. Completing all eligible HHAR and PAMAP2 outer-person rotations materially changes the conclusion obtained from the original frozen fold. On HHAR, balanced accuracy on that single person is 95.6-97.2% across methods versus 78.3-83.0% over all nine users, a reduction of 13.8-17.8 percentage points (pp). The displayed mean leader changes on both datasets, while paired leader-runner bootstrap intervals include zero and do not resolve a superior method. No adaptive core mechanism combines positive mean gain in all five datasets with zero seed-averaged person-level losses greater than 2 percentage points (pp). FedBN has one such loss and ATP-style adaptation has eight; Feature-only has none after seed averaging, but its exact one-sided 95% upper bound is 7.6%. A complementary seed-person stress audit records 4, 22, and 10 harmful realizations out of 114 for FedBN, ATP-style, and Feature-only, respectively; these are repeated realizations, not independent participants. Tail quality, calibration availability, and fall-window specificity reveal additional failures hidden by mean accuracy. The study therefore provides an auditable development benchmark and failure map rather than a universal-superiority or deployment-safety claim.
Sources
- Federated Learning with Personalization Layers
- BayaHAR: Lightweight Bayesian Few-Shot User Adaptation for On-Device Personalized Human Activity Recognition
- Temporal Structure Matters for Efficient Test-Time Adaptation in Wearable Human Activity Recognition
- PI-TTA: Physics-Informed Source-Free Test-Time Adaptation for Robust Human Activity Recognition on Mobile Devices
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks