- This event has passed.
All that likelihood with PyMC3 by Dr. Junpeng Lao
11/06/2018 @ 19:00 - 21:00
Event will be in English, as always.
Location: Advaneo, Neuer Zollhof 2, Düsseldorf
Title: All that likelihood with PyMC3
Speaker: Dr. Junpeng Lao
The likelihood is a central concept in Bayesian computation. In this tutorial, we will learn about what is the likelihood function and how do we use it for inference. Using PyMC3, I will demonstrate how to get the likelihood from a model, how does it connect to inference using NUTS or Variational approximation, and some practical usage of the model likelihood to perform model comparisons.
In the past years, we see a growing interested in probabilistic programming and Bayesian statistics. While modern probabilistic library such as PyMC3 and Stan provide flexibility to the user to write down complex model easily, it is not always intuitive how inference is done. More specifically, what is the sampler “seeing”? Learning about likelihood is central to the understanding of how inference is performed. In this tutorial, I would like to give some introduction to likelihood function and why they are important, what does mean to put constraints in the likelihood, how to get likelihood from PyMC3 and what is the correspondent input.
I am Junpeng Lao, currently a postdoc in the Department of Psychology at the University of Fribourg.
With a background in Psychology (B.Sc. from Sun Yat-sen University and PhD from the University of Glasgow) and a deep interested in Mathematics and Computation, I study the computational nature of the neural system using various techniques including psychophysics, eye tracking, EEG, fMRI, and computation models. I am very passionate about Statistics and Scientific Computing. Currently, I am a member of pymc-devs and a regular contributor to PyMC3. I also maintain the PyMC3 user community.
You can find out more about me on my website: http://junpenglao.xyz/