Patryk Marszałek

ICED achieves competitive performance across multiple density estimation tasks without the need for retraining or hyperparameter tuning, revolutionizing the efficiency of tabular data analysis.

Vision-language models aren't as entangled as we thought: a simple rotation can reveal surprisingly clean, interpretable features without adding parameters.

CounterFlowNet generates more diverse and plausible counterfactual explanations than existing methods by framing the search as a sequential decision process learned by a GFlowNet.