Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
1
2
1
LPF offers provable guarantees on calibration, robustness, and uncertainty decomposition in multi-evidence reasoning, addressing a critical gap in trustworthy AI.
Achieve 97.8% accuracy and 1.4% calibration error in multi-evidence reasoning by fusing VAE latent uncertainty with Sum-Product Networks, outperforming LLMs and other baselines.