Probabilistic Graphical Models (week8)

Learning Tree-structured Networks

In the past few weeks, you have learned about parameter estimation in probabilistic graphical
models, as well as structure learning. In this programming assignment, you will explore structure
learning in probabilistic graphical models from a synthetic dataset. In particular, we will provide
you synthetic human and alien body pose data. We use a directed model to represent their body
poses. You will tackle problems including learning CPDs for continuous variables and learning
tree-structured graphs among body parts. Finally, you will use the learned models to classify
unseen pose data as either humans or aliens.


倒数第二周！

这周已经各种晕了，感觉自从learning as inference之后就一直没有完整理解思维体系。作业做起来即使能得到结果也仍旧不知所云= =

做完之后反复想，结合第九次编程作业，可以理解不少。不过用线性高斯模型对x, y和alpha一起建模合适吗？