Occupancy estimation and activity recognition in smart buildings using open set domain adaptation
Bibliographic record
Abstract
Smart buildings (SB) optimize energy use to help the world achieve better sustainability while supporting energy efficiency goals and reversing climate change. Successful occupancy estimation (OE) along with activity recognition (AR) provides essential functionality to accomplish these objectives through the modification of building systems (BS), including heating and lighting. However, developing robust models for SB tasks is challenged by the scarcity of labeled data. Thus, the need for domain adaptation, that aims at adapting a model trained on one domain (source) to perform well on a different domain (target). Traditional closed-set domain adaptation (CSDA) methods have a critical limitation: they assume shared classes between source and target domains, which is unrealistic in many applications. Open Set Domain Adaptation (OSDA) provides a more realistic solution, assuming that only a few categories are available in both domains, making OSDA methods more suitable for SB. This paper presents four novel OSDA approaches: OSDA by backpropagation, OSDA with soft rejection, Unknown Aware Domain Adversarial Learning (UADAL) for OSDA, and Adjustment and Alignment (ANNA) for unbiased OSDA, to address these challenges. Evaluated on offices and residential apartment datasets, our top-performing method achieves 92% accuracy, a 0.89 F1-score on known classes, and an 85% unknown-class detection rate. These techniques differentiate between known and unknown target classes through adversarial training, reducing negative transfer, and improving feature alignment. The OSDA methodologies show notable improvements through testing on SB datasets. This indicates that OSDA implementation in SB can result in significant energy savings and improved sustainability. The code is available in this repository: https://github.com/ons-abderrahim/OSDA-to-Improve-Occupancy-Estimation-and-Activity-Recognition-in-Smart-Buildings .
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 | 0.002 |
| 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 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".