João Batista, former PhD student supervised by Sara Silva, published the paper entitled “Complexity, interpretability and robustness of GP-based feature engineering in remote sensing” in the journal “Swarm and Evolutionary Computation”, ranked within the top 10% of the Scimago Journal Ranking in different categories and ranked 4th by Google Scholar in the Evolutionary Computation category.

The paper analyses the feature engineering needs of different classifiers and proposes new models for detecting cocoa agroforest and forecasting forest degradation with remote sensing data. It also studies the correlation between model complexity and predictive ability, proposing a new functional complexity metric for classification datasets.