20241010 - TA - PLSA algorithm

The meeting discussed various algorithms and models such as Latent Semantic Analysis, Probability LSA, and the EM algorithm, focusing on refining parameters and achieving convergence in topic modeling.
Latent semantic analysis with singular value composition is a known algorithm.
Probability LSA is a probability model that explains the creation of documents using distribution.
The EM algorithm is used in the clustering algorithm, with an e-step to find expected values and an m-step to find maximum values.
The next step is to refine the P(w|z) and P(d|z) parameters through the E step and M step until convergence.
Convergence is reached when there are no significant changes in the model parameters or likelihood of observed data.
Soft assignment is used in topic modeling, where words may belong to multiple topics with varying probabilities.