Federated optimal transport domain adaptation using Internet of Things sensor data for activity recognition
Bibliographic record
Abstract
Federated learning has recently gained interest in optimizing smart buildings by addressing privacy issues. Activity recognition (AR) is important in improving smart building applications using Internet of Things (IoT) sensor data, allowing HVAC systems optimization, energy efficiency, security, and occupant comfort. However, the development of AR applications is often limited due to labeled data scarcity. IoT data is difficult to collect due to privacy, cost, and time issues. Common unsupervised domain adaptation (UDA) methods address these issues by sharing information from labeled sources to unlabeled target domains. Most of these approaches require direct access to labeled source data, raising privacy issues. This research proposes novel approaches that integrate federated learning (FL) with optimal transport domain adaptation (OTDA) to perform activity recognition while respecting privacy concerns. Optimal transport (OT)-based approaches have gained much interest thanks to their effective probability distribution alignment and improved generalization between source and target data features. Also, our novel methods use federated learning to allow the training of several models from multiple smart building sources without sharing the local data of each source domain. FL creates a robust global model that combines the performance of several local models to perform AR. This research paper introduces a novel integration of federated learning and optimal transport domain adaptation to achieve privacy-preserving and activity recognition across heterogeneous domains, which represents a new contribution to smart building research. We have developed two federated learning methods (federated averaging (FedAvg) and federated proximal (FedProx)) with optimal transport domain adaptation (FL-OTDA) called FedAvg-OTDA and FedProx-OTDA. Experimental results of global models on unseen AR datasets, with scores up to 71%, prove that the developed FL-OTDA methods are promising for the current smart building task. We provide the code in the following repository: https://github.com/JawDri/Federated-Learning-Optimal-Transport-Domain-Adaptation-for-Activity-Recognition.git .
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.004 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".