Search papers, labs, and topics across Lattice.
Affiliation:
5
0
6
Memory contamination can mislead performance metrics in anomaly detection, revealing that less contaminated memories don't always yield better results.
CLEANCON achieves near-zero memory contamination while paradoxically showing that less contamination doesn't guarantee better anomaly detection performance.
ProCon achieves unprecedented anomaly detection accuracy without the need for training or pseudo-anomaly supervision, redefining the capabilities of memory-based methods.
A 1000x larger video reasoning dataset reveals early signs of emergent generalization, offering a new foundation for training and evaluating spatiotemporal AI.
Ditch max pooling for anomaly detection: StructCore unlocks near-perfect image-level AUROC scores (99.6% on MVTec AD) by analyzing the *structure* of anomaly score maps, all without any training.
