[Coursera] Probabilistic Graphical Models (Week #4)

previously:
第三周   http://www.1point3acres.com/bbs/thread-40990-1-1.html
第二周   http://www.1point3acres.com/bbs/thread-40349-1-1.html
第一周   http://www.1point3acres.com/bbs/thread-39882-1-1.html

Stanford CS228: Probabilistic graphical models WEEK 4

Inference: Belief Propagation
    Properties of Belief Propagation
    Clique Tree Algorithm
    Clique Trees and Independence
    Clique Trees and Variables Elimination
    Loopy BP and Message Decoding
Inference: MAP Estimation
    Max Sum Message Passing
    Finding a MAP Assignment

第四次作业
In PA-1, you implemented a rudimentary inference engine that could correctly answer queries over small networks. However, the “brute force” inference engine that you implemented in PA-1 is unable to handle anything larger than tiny networks; its running time is proportional to the number of entries in the joint distribution over the entire network.
Instead of "brute force" method, We will explore "clique tree message passing" in this Assignment, and by its end, you will have created an inference engine powerful enough to handle probabilistic queries and find MAP assignments over the "genetic inheritance networks" dataset from PA-2 and the "OCR networks" dataset from PA-3, respectively.

第四次作业

求学分求学分

这次没做满。。

求问5~9周的都在哪？没人发我过两天去起个帖子~话说地里没人上这个课么？