Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
School of Mechanical and Material Engineering, Department of Mechanical Engineering, University College Dublin, Massachusets Institute of Technology
3
0
3
Embedding scientific principles into training distributions can drastically improve the accuracy and interpretability of latent representations in complex hypothesis spaces.
Length-adjusted tail entropy can serve as a powerful signal for solution quality, enabling significant performance boosts in inference-time scaling across complex domains.
Random train/test splits hide critical weaknesses in FOWT fatigue prediction models, but FLOATBench's new regime-aware benchmark exposes extrapolation failures that could lead to catastrophic structural failures.