Metraipolis › archive › dc_3c5cb0240147

SAE-Xplainers: Rule-Based Feature Interpretation for Extreme Earth Events — arXiv:2608.20117

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.20117
License (as archived)
CC0-1.0
Archived
2026-10-18T11:48:25.570Z
SHA-256 of the archived text
08296810a61ec3e9b0c95e36ae503f3aa75b0ca0fd7131e5f52ddd16dd32e6e2
Stable URL of this page
https://metraipolis.com/documento/dc_3c5cb0240147

How to cite

SAE-Xplainers: Rule-Based Feature Interpretation for Extreme Earth Events — arXiv:2608.20117 — arXiv (en), https://arxiv.org/abs/2608.20117 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3c5cb0240147 (Metraipolis archive dc_3c5cb0240147, sha256 08296810a61ec3e9b0c95e36ae503f3aa75b0ca0fd7131e5f52ddd16dd32e6e2)
BibTeX
@misc{dc3c5cb0240147,
  title = {SAE-Xplainers: Rule-Based Feature Interpretation for Extreme Earth Events — arXiv:2608.20117},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.20117},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3c5cb0240147; sha256 08296810a61ec3e9b0c95e36ae503f3aa75b0ca0fd7131e5f52ddd16dd32e6e2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.20117. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The emergence of large-scale Weather and Climate (W&C) datasets offers new opportunities for modeling extreme Earth events (ExEE) and their impacts using deep learning. However, their adoption in operational settings remains limited by the lack of models' interpretability. While for conventional text and image modalities, tools such as Sparse Autoencoders (SAEs) have proven effective for extracting human-understandable concepts, their use for the analysis of ExEE remains challenging due to the nature of W&C data. To address this, we introduce (i) a geographic location-based modulation of the inputs of SAE to capture the local semantic meaning of environmental patterns, and (ii) an ensemble of rule-based SAE-Xplainers to interpret the resulting high-dimensional features derived from complex, multi-modal environmental predictors. We evaluate our method on three ExEE types: the prediction of fires, and the detection of tropical cyclones and atmospheric rivers. We show that SAE input modulation improves both reconstruction performance and feature utilization, and that our SAE-Xplainers enable faithful interpretation of complex climatic patterns by unfolding them into human-understandable rules that are consistent with the scientific literature, while also supporting the identification of feature absorption.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-21 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.20117 — 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 2026-10-18T11:47:43.337Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).