MIRANDA: MI d-feature RAN k-adversarial D omain A daptation toward climate change-robust ecological forecasting with deep learning

Yuchang Jiang [email protected] Jan Dirk Wegner [email protected], DML, UZH, Zurich, SwitzerlandVivien Sainte Fare Garnot [email protected] · 2026

Read MIRANDA: MI d-feature RAN k-adversarial D omain A daptation toward climate change-robust ecological forecasting with deep learning by Yuchang Jiang [email protected] Jan Dirk Wegner [email protected], DML, UZH, Zurich, SwitzerlandVivien Sainte Fare Garnot [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-00800.txt.

Description

Jiang, Wegner, and Sainte Fare Garnot introduce MIRANDA, a domain-adaptation method that keeps plant phenology forecasts robust as climate shifts. Instead of forcing invariance on final features, it applies rank-based adversarial training to middle features. That enforces year-invariance in learned weather representations. Tested on 70 years of data with 67,800 observations across 5 tree species, MIRANDA narrows the gap with mechanistic models, unlike standard domain-adaptation methods.

Catalog record

CREATORYuchang Jiang [email protected] Jan Dirk Wegner [email protected], DML, UZH, Zurich, SwitzerlandVivien Sainte Fare Garnot [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE44877
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.00800

Browse the full library · About the Tower of Babel