Anthill Inside 2019

##Details of the workshop (date, time, venue) and tickets available here: https://hasgeek.com/anthillinside/bayesian-networks-tutorial/

Bayesian networks (BNs) are graphical structures that capture the probabilistic relationships between several random variables. They are a natural fit for scenarios that can benefit from both causal as well as probabilistic semantics, thereby, gracefully combining prior knowledge (in causal form) and knowledge from data. In this tutorial, we discuss methods to construct Bayesian networks from prior knowledge and statistical methods to improve these models through data. We will cover the exact and approximate inference techniques as well as techniques to learn the parameters and structure of BNs. We illustrate the BN modeling approach using real-world case study.

Outline

30 mins
Introduction to Bayesian Networks

• Recap of probability theory, random variables
• Conditional probability and independence
• Bayes theorem
• Joint probability distributions
• Markov condition
• Bayesian networks

45 mins
Inference in Bayesian Networks

• Message passing algorithm
• The Noisy-OR model
• Variable elimination
• Continuous variable inference
• Approximate inference techniques

45 mins
Parameter learning

• Learning Parameters in a Bayesian Network
• Learning with Missing Data Items
• Multinomial Variables
• Continuous Variables

45 mins
Bayesian structure learning

• Learning Structure: Discrete Variables
• Model Averaging
• Learning Structure with Missing Data
• Probabilistic Model Selection
• Learning Structure: Continuous Variables

45 mins
Applications

Speaker bio

Ashish Kulkarni has over 10 years of industry experience and currently heads the Data Science Lab at Clustr. He holds a Ph.D. in Computer Science from IIT Bombay and has published papers in top tier conferences lile AAAI, IJCAI and others. Prior to joining Clustr, he worked as an applied machine learning scientist at Amazon.

Links

• https://www.linkedin.com/in/ashishakulkarni/