Timezone: »

Bayesian Models of Data Streams with Hierarchical Power Priors
Andres Masegosa · Thomas D. Nielsen · Helge Langseth · Dario Ramos-Lopez · Antonio Salmeron · Anders Madsen

Mon Aug 07 01:30 AM -- 05:00 AM (PDT) @ Gallery #37

Making inferences from data streams is a pervasive problem in many modern data analysis applications. But it requires to address the problem of continuous model updating, and adapt to changes or drifts in the underlying data generating distribution. In this paper, we approach these problems from a Bayesian perspective covering general conjugate exponential models. Our proposal makes use of non-conjugate hierarchical priors to explicitly model temporal changes of the model parameters. We also derive a novel variational inference scheme which overcomes the use of non-conjugate priors while maintaining the computational efficiency of variational methods over conjugate models. The approach is validated on three real data sets over three latent variable models.

Author Information

Andres Masegosa (University of Almeria)
Thomas D. Nielsen (Aalborg University)
Helge Langseth (Norwegian University of Science and Technology)
Dario Ramos-Lopez (University of Almeria)
Antonio Salmeron (University of Almeria)
Anders Madsen (Hugin Expert A/S)

Related Events (a corresponding poster, oral, or spotlight)