Current browse context:
cs
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data
(Submitted on 20 May 2016 (v1), last revised 3 Mar 2017 (this version, v3))
Abstract: We introduce Deep Variational Bayes Filters (DVBF), a new method for unsupervised learning and identification of latent Markovian state space models. Leveraging recent advances in Stochastic Gradient Variational Bayes, DVBF can overcome intractable inference distributions via variational inference. Thus, it can handle highly nonlinear input data with temporal and spatial dependencies such as image sequences without domain knowledge. Our experiments show that enabling backpropagation through transitions enforces state space assumptions and significantly improves information content of the latent embedding. This also enables realistic long-term prediction.
Submission history
From: Maximilian Soelch [view email][v1] Fri, 20 May 2016 16:52:22 GMT (2268kb,D)
[v2] Sat, 23 Jul 2016 07:33:14 GMT (6722kb,D)
[v3] Fri, 3 Mar 2017 18:12:53 GMT (7048kb,D)
Link back to: arXiv, form interface, contact.