From: Stepwise iterative maximum likelihood clustering approach
1. Initialization: select initial partitions with means μ 1, μ 2, …, μ c and covariance matrices Σ 1, Σ 2, …, Σ c | |
2. Loop: Select a sample \( \widehat{\mathbf{x}}\in {\chi}_i \). | |
3. If n i > 1 then compute | |
4. \( {\delta}_j=\left\{\begin{array}{c}\hfill \varDelta {L}_j,\kern0.75em j\ne i\hfill \\ {}\hfill \varDelta {L}_i,\kern0.75em j=i\hfill \end{array}\right. \) | |
5. Transfer \( \widehat{\mathbf{x}} \) to χ k if δ k = max δ j for all j. | |
6. Update L tot , μ i , μ k , Σ i and Σ k . | |
7. If L tot doesn’t change in n attempts then stop otherwise go to Loop. |