Jonas Peters

Anchor PCA reveals that focusing on shared variation across domains can significantly enhance the robustness of unsupervised dimension reduction, outperforming traditional pooling methods.

By cleverly mining historical data for invariances, ISD-linUCB achieves superior regret in non-stationary bandits, even when the reward model changes rapidly.

Unlock domain generalization with unlabeled data by exploiting the structure of anti-causal relationships, where outcomes cause covariates.