Imitative Non-Autoregressive Modeling for Trajectory Forecasting and Imputation

CVPR(2020)

引用 45|浏览241
暂无评分
摘要
Trajectory forecasting and imputation are pivotal steps towards understanding the movement of human and objects, which are quite challenging since the future trajectories and missing values in a temporal sequence are full of uncertainties, and the spatial-temporally contextual correlation is hard to model. Yet, the relevance between sequence prediction and imputation is disregarded by existing approaches. To this end, we propose a novel imitative non-autoregressive modeling method to simultaneously handle the trajectory prediction task and the missing value imputation task. Specifically, our framework adopts an imitation learning paradigm, which contains a recurrent conditional variational autoencoder (RC-VAE) as a demonstrator, and a non-autoregressive transformation model (NART) as a learner. By jointly optimizing the two models, RC-VAE can predict the future trajectory and capture the temporal relationship in the sequence to supervise the NART learner. As a result, NART learns from the demonstrator and imputes the missing value in a non autoregressive strategy. We conduct extensive experiments on three popular datasets, and the results show that our model achieves state-of-the-art performance across all the datasets.
更多
查看译文
关键词
temporal relationship,NART learner,nonautoregressive strategy,trajectory forecasting,pivotal steps,human objects,temporal sequence,spatial-temporally contextual correlation,sequence prediction,trajectory prediction task,missing value imputation task,imitation learning paradigm,recurrent conditional variational autoencoder,RC-VAE,nonautoregressive transformation model,imitative nonautoregressive modeling
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要