http://www.columbia.edu/~jwp2128/Papers/ZhangPaisley2018.pdf
The paper introduces a novel method to integrate Bayesian nonparametrics and deep neural networks for time-series data. It extends linear Gaussian basic model to a neural network and use the variational auto-encoder for approximate posterior inference. Accroding to this:
Project goal:
- implement the proposed method
- repeat the authors experiments
- apply to the new data
- Rasul Khasyanov
- Alexander Parubchenko