Search papers, labs, and topics across Lattice.
2
0
3
Scaling generalized zero-shot recognition to massive vocabularies does not require re-architecting backbones: treating seen-versus-unseen gating as a downstream Monte Carlo calibration problem boosts unseen accuracy by over 20% in just 15 dimensions.
Balanced datasets can dramatically improve emotion recognition performance, especially for minority emotions often neglected in existing benchmarks.
