a probabilistic latent semantic analysis model in matlab programming
1 introduction
I have implemented the probabilistic latent semantic analysis model in matlab, plus with a runnable demo. On one hand, it is used to make myself further familar with the plsa inference. On the other hand, it is very interesting to do programming in matlab.
In this implementation, it is basically consistent with original paper. Besides, I also refer to the below document,
[1] plsa and em algorithm, http://blog.csdn.net/yangliuy/article/details/8330640
[2] EM algorithm, http://cs229.stanford.edu/notes/cs229-notes8.pdf
[3] A not on plsa and em, http://luxinxin.is-programmer.com/user_files/luxinxin/File/plsanote.pdf
[4] em algorithm wiki, http://en.wikipedia.org/wiki/Expectation%E2%80%93maximization_algorithm
For em argorithm in plsa, two derived approches can be done. One has been described in [1] and [3]. The other can been seen in [2]. For more information on EM, please refer to [4].
In the next, I plan to make slides to give a introduction to PLSA model in terms of resources at hand. It will be shared with you.
2 Conclusion
It is simple implementation. I am not sure whether there are other potential bugs in it. If you have any feedback, Please let me know.