Search papers, labs, and topics across Lattice.
University of Georgia
3
0
5
Pretrained time-series foundation models offer surprisingly inconsistent benefits for EMG tasks, with full fine-tuning often outperforming linear probing and few-shot adaptation only partially recovering performance on new users.
Meta-optimizing an AI data scientist can dramatically enhance the quality of synthetic datasets, outperforming traditional methods.
Urban trajectory generation is a multi-objective challenge where no single model can dominate all performance metrics, revealing critical trade-offs in generative approaches.
