Seminar on Probabilistic Programming

Algorithms and machine learning
Advanced studies
Year Semester Date Period Language In charge
2016 autumn 09.09-16.12. 1-2 English Antti Honkela


Time Room Lecturer Date
Fri 10-12 C220 Antti Honkela 09.09.2016-21.10.2016
Fri 10-12 C220 Antti Honkela 04.11.2016-16.12.2016


Probabilistic models are popular tools for data analysis and machine learning, but applying them has traditionally required either using specific pre-implemented models or manually deriving and implementing inference algorithms, which is laborious and requires expertise. Probabilistic programming aims to make it easier to apply different models by providing tools that automatically perform inference given a description of the model. From a computer science perspective, these tools “compile” models into inference algorithms.
Probabilistic programming covers a broad range of tools from ones that perform inference over a relatively restricted set of models to ones that allow specifying arbitrary computations on random variables using a Turing complete language. The field is currently advancing very quickly, thanks to advances in inference algorithms for “black box” models. A recent paper “Human-level concept learning through probabilistic program induction” demonstrated how probabilistic programming approaches can learn complex concepts significantly more efficiently than state-of-the-art deep neural network approaches.

This seminar looks at the probabilistic programming concept covering both the basic approaches and the recent advances, including also the very latest scientific articles and recently published tools. During the seminar you will see several probabilistic programming tools that can be applied for solving practical modeling tasks.


The seminar is given by Academy Research Fellows Antti Honkela and Arto Klami.



NEWS: The course has a Moodle page with further instructions and material.


Completing the course

Each participant will study in detail 1-3 research or review articles about the topic (article suggestions will be provided) and give an oral presentation (in Period II) for other participants. In addition, everyone will write a 4-6 page report on the topics covered during the seminar, review the reports of some other participants, and act as an opponent for the oral presentations.



  • Sep 9: Introduction
  • Sep 16: Practical instructions
  • Oct 7: 5 minute initial presentations
  • Nov 4: Deadline for the written report
  • Nov 11 - Dec 2: Seminar presentations (details in a separate tab)
  • Nov 18: Deadline for the reviews
  • Dec 2: Deadline for the revised reports


Topics are listed in a separate tab.