Can I read Learning Temporally Causal Processes on EtoBox?

Learning Temporally Causal Processes by Yewei Xia is a document available to read on EtoBox.

What is Learning Temporally Causal Processes about?

This paper proposes a new framework called LEAP to learn temporally causal latent processes from general temporal data in an unsupervised manner. The paper establishes conditions under which temporally causal latent processes can be identified from their nonlinear mixtures without assumptions of sparsity or minimality. LEAP extends variational autoencoders by enforcing these conditions through constraints in a causal process prior network. Experimental results demonstrate LEAP can reliably identify temporal

Author

Yewei Xia

Language

EN