Kristoffer Wickstrøm

Neural networks can learn interpretable, polytope-based latent representations that preserve performance and enable direct VQ training.

LMV-Net outperforms existing breast cancer risk prediction models by effectively integrating longitudinal multi-view data, revealing the critical role of spatial-temporal information in enhancing risk stratification.

Anomaly detection just got a major upgrade: ProtoX-AD delivers high accuracy with explanations that make sense, bridging the gap between performance and interpretability.