Search papers, labs, and topics across Lattice.
Northeastern University London
2
0
4
Bridging the gap between coarse atmospheric models and local PM$_{2.5}$ variations, this framework achieves a 40x super-resolution without relying on temporal data.
Rectified flows can generate synthetic skin lesion images that boost classification accuracy by up to 9% compared to diffusion models, offering a promising solution to data scarcity in dermatology.
