Search papers, labs, and topics across Lattice.
3
0
3
AdvScene, a conditional latent diffusion model that is trained in two stages that Conditioning on ego driving displacement prevents the ego from remaining static, and RL finetuning induces criticality directly with non-differentiable safety-critical metrics is proposed.
Model-level mixtures of experts can significantly boost object detection performance while enhancing interpretability, outperforming conventional ensemble methods.
Patch-wise Mixture-of-Experts can boost CNN semantic segmentation performance by nearly 4 mIoU, but only if you carefully tune the architecture.
