Abstract:
The aim of the online course Bayesian and Causal Bayesian Networks is to introduce the theory and provide the necessary skills to apply these machine-learning models in practice. With the revelation of artificial intelligence and machine learning models, the world has witnessed an increasing desire to use them in different applications. An obstacle preventing the wide use of machine learning models is their "black box" nature -- a quality referred to as "uninterpretable". Classical mechanistic models that are based on our prior understanding of the world are often trusted and preferred, but they often fall short in performance. Bayesian networks, which are probabilistic graphical models, nicely fill in this gap, as they are graphical, and hence, relatively easy to understand, yet as powerful as advanced machine learning models. Moreover, they have been recently extended to causal Bayesian networks to systematically identify causal relationships in unknown processes, making them more intuitive and reliable. In addition to recorded lectures, this course includes several micro instructional videos that break the heavy material into small digestible pieces. The micro videos allow a broader range of audience with limited time and background knowledge to benefit from this course.