A Sparse and Adaptive Prior for Time-Dependent Model Parameters
This work addresses the challenge of handling time-dependent parameters in probabilistic modeling, particularly for financial and language applications, but appears incremental as it builds on existing variational inference methods.
The authors tackled the problem of modeling time-varying parameters in probabilistic models by introducing a novel prior that promotes sparsity and adapts correlation strength based on data, achieving results in forecasting financial quantities from text and modeling language with time-varying financial measurements.
We consider the scenario where the parameters of a probabilistic model are expected to vary over time. We construct a novel prior distribution that promotes sparsity and adapts the strength of correlation between parameters at successive timesteps, based on the data. We derive approximate variational inference procedures for learning and prediction with this prior. We test the approach on two tasks: forecasting financial quantities from relevant text, and modeling language contingent on time-varying financial measurements.