Tag: Shapley values

Researchers developed a Shapley-guided token pruning method that significantly reduces latency and energy consumption in Vision Transformers for remote sensing ...

A new diagnostic framework applied to Dutch health records reveals that standard fairness metrics can hide feature-level patterns of algorithmic ...

A new inferential framework tests whether entire groups of predictors significantly shape machine learning predictions, revealing that market conditions and ...

Explainable AI applied to nearly 2,000 mammal species reveals that climate extremes, temperature trends, and land change exert sharply divergent ...

A new explainable AI pipeline converts breast histology classifiers' evidence into auditable, region-level attribution maps and fully reproducible SOAP-style clinical ...

A new study argues that criminologists have been testing the liberation hypothesis with regression effects when the theory actually concerns ...

Researchers have developed a rough-set-based AI framework that quantifies Traditional Chinese Medicine diagnostics and fuses them with Western clinical data, ...

A Stanford-led team trained supervised machine learning models on nationally representative EHR-like survey data to identify people with untreated alcohol, ...

Researchers at the Norwegian University of Science and Technology have developed two new explainability methods that integrate transformer attention weights ...

Researchers in Porto have developed a game theory-inspired machine learning framework that gives each transportation mode its own tailored feature ...