search

ZikangZhou / HiVT

[CVPR 2022] HiVT: Hierarchical Vector Transformer for Multi-Agent Motion Prediction
https://openaccess.thecvf.com/content/CVPR2022/papers/Zhou_HiVT_Hierarchical_Vector_Transformer_for_Multi-Agent_Motion_Prediction_CVPR_2022_paper.pdf
Apache License 2.0
577 stars 115 forks source link

question regarding Laplace distribution #24

Closed dichencd closed 1 year ago

dichencd commented 1 year ago

Hi @ZikangZhou, when calculating regression loss, you seem to model x and y as independent univariate Laplace distributions if I understand correctly based on your code. I wonder what would be the intuition behind such a choice versus considering x & y jointly as a multivariate Laplace distribution? Thank you very much for your comments in advance!

ZikangZhou commented 1 year ago

Hi,

Thanks for your interest! I would say it's for simplicity, although this assumption is somewhat counter-intuitive. You can try to model x & y jointly. I think many people are interested in such an ablation study!

dichencd commented 1 year ago

I see. Thanks!