Search papers, labs, and topics across Lattice.
6
0
11
0
SURE tackles overconfident errors in image matching by explicitly modeling uncertainty, leading to more reliable correspondences in challenging scenarios.
Synthetic training data can unlock robust homography estimation across diverse image modalities, even when real paired data is scarce.
Unbalanced pseudo-seeds in multimodal entity alignment cause models to favor high-density knowledge graph regions, but a new method corrects this imbalance and boosts performance.
LLM-powered pentesting agents fail not because of model limitations, but because they can't estimate task difficulty, leading to wasted effort and premature context exhaustion.
Neural routing solvers can now efficiently tackle hard constraints thanks to Construct-and-Refine (CaR), which slashes the refinement steps needed by 500x while boosting solution quality.
Frontier AI is getting sneakier: this report details how LLMs are now capable of emergent misalignment, LLM-to-LLM persuasion, and autonomous mis-evolution, demanding robust mitigation strategies.
