Bidirectional Information Flow enables continuous two-way communication in hierarchical Gaussian processes for Bayesian optimization, improving sample efficiency, training robustness, and modular subtask reuse while significantly outperforming unidirectional and vanilla methods.
IMBUE enables amortized Bayesian experimental design to incorporate external deployment knowledge via in-context tokens and a reliability filter, accelerating early information gain with reliable inputs while maintaining baseline performance otherwise.
PPAT combines unbiased LURE estimation with prediction-powered control variates and adaptive acquisition to reduce label variance, yielding valid confidence intervals with fewer labels.
Strategic decision-focused learning predicts exogenous states for multi-agent games where better accuracy can reduce equilibrium payoffs, requiring strategic-aware predictors.
A geospatial discovery framework combines active learning, online meta-learning, and concept relevance to robustly find hidden targets like PFAS under sparse, changing conditions with limited sampling budgets.