FedGraph: Probing-Based Personalized Federated Learning for Human Activity Recognition with Multimodal Physiological Signals
Bibliographic record
Abstract
Traditional human activity recognition (HAR) models using federated learning (FL) with inertial sensors often struggle to differentiate between activities with subtle physical differences, such as distinguishing between sitting while driving and sitting while having lunch. This paper introduces FedGraph, a personalized FL approach that enhances HAR by incorporating multimodal physiological signals, including photoplethysmogram (PPG) and electrodermal activity (EDA), alongside inertial measurements. FedGraph employs a probing-based graph filtering mechanism to personalize model aggregation without relying on sensitive metadata such as user demographics or hardware specifications. Unlike conventional approaches, FedGraph dynamically constructs a model similarity graph by probing local models with synthetic signals, generating latent embeddings that capture client-specific characteristics. This probing approach eliminates the need for explicit metadata, allowing FedGraph to adaptively personalize models in heterogeneous environments while preserving privacy. We evaluate FedGraph on the DaLiA dataset, which includes multimodal physiological and inertial signals from 15 subjects performing daily activities. Experimental results demonstrate that FedGraph significantly outperforms both the vanilla FedAvg and metadata-driven FedGraph variants, confirming its effectiveness in improving model personalization and robustness while ensuring privacy preservation in a federated setting.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".