Metraipolis › archive › dc_188ee807713f
Shift Aware Transfer Learning with Adaptive Dual-Encoder Fusion for PM Forecasting in Data-Limited Environments — arXiv:2608.14456
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.14456
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T02:20:51.721Z
- SHA-256 of the archived text
05520d5e09b999722fba1fd1dfc94f62151b67b598a17137c9d8cb0027671abd- Stable URL of this page
https://metraipolis.com/documento/dc_188ee807713f
How to cite
Shift Aware Transfer Learning with Adaptive Dual-Encoder Fusion for PM Forecasting in Data-Limited Environments — arXiv:2608.14456 — arXiv (en), https://arxiv.org/abs/2608.14456 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_188ee807713f (Metraipolis archive dc_188ee807713f, sha256 05520d5e09b999722fba1fd1dfc94f62151b67b598a17137c9d8cb0027671abd)
BibTeX
@misc{dc188ee807713f,
title = {Shift Aware Transfer Learning with Adaptive Dual-Encoder Fusion for PM Forecasting in Data-Limited Environments — arXiv:2608.14456},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.14456},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_188ee807713f; sha256 05520d5e09b999722fba1fd1dfc94f62151b67b598a17137c9d8cb0027671abd},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.14456. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Short-horizon forecasting of fine particulate matter (PM2.5) remains difficult when observations from the target domain are limited and the statistical properties of the source and target domains differ. In these settings, models trained only on local data may not capture complex temporal dynamics, while direct transfer learning can result in negative transfer. This study develops a shift-aware dual-encoder transfer framework that combines source-domain knowledge with target-specific representation learning. The source encoder was pretrained using hourly observations from 10 U.S. monitoring locations. The framework was then adapted and evaluated using two years of hourly observations from 77 stations in Taiwan under a chronological train-validation-test protocol. Among the four principal baselines, the frozen-source dual-encoder model achieved the best performance, with MSE = 21.8960, MAE = 3.1597, and R^2 = 0.8725. This corresponds to an MSE reduction of approximately 7.1% relative to TL-v1 and 4.1% relative to TL-v2. The ablation analysis showed that removing the Taiwan-specific branch caused the largest decline in performance. Allowing the source encoder to adapt produced the best overall result, with MSE = 21.6575, MAE = 3.1383, and R^2 = 0.8739. SHAP analysis indicated that predictions were driven mainly by recent PM2.5 observations and meteorological variables related to pollutant transport and dispersion. These results suggest that source-domain knowledge is most effective when target-specific information is preserved and the transferred representation is allowed to adapt under target supervision. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-17 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.14456 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-19T02:02:02.825Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).