1. The document discusses Edward, a Python library for probabilistic modeling, inference, and criticism built on top of TensorFlow. It combines TensorFlow for computation and additional probabilistic programming languages (PPLs) for probabilistic modeling.
2. It provides an overview of TensorFlow's key capabilities like GPU/TPU support and high-level APIs and compares it to PPLs' abilities for probabilistic modeling using distributions and performing inference using techniques like variational inference and MCMC.
3. Edward allows building probabilistic models with TensorFlow and performing inference using techniques from PPLs to take advantage of both frameworks' strengths. This allows tasks like Bayesian neural networks and deep generative models.
29. Edward
- Edward = TensorFlow + +
- TensorFlow
-
- TF GPU, TPU, TensorBoard, Keras
-
- Box’s Loop
- Python
31. Refrence
•D. Tran, A. Kucukelbir, A. Dieng, M. Rudolph, D. Liang, and D.M.
Blei. Edward: A library for probabilistic modeling, inference,
and criticism.(arXiv preprint arXiv:1610.09787)
•D. Tran, M.D. Hoffman, R.A. Saurous, E. Brevdo, K. Murphy, and
D.M. Blei. Deep probabilistic programming.(arXiv preprint
arXiv:1701.03757)
•Box, G. E. (1976). Science and statistics. (Journal of the
American Statistical Association, 71(356), 791–799.)
•D.M. Blei. Build, Compute, Critique, Repeat: Data Analysis with
Latent Variable Models. (Annual Review of Statistics and Its
Application Volume 1, 2014)