I have implemented the Von Mises distribution (VM) as another continuous distribution. It is structured in the same way as the Gaussian Distribution in terms of using the same functions for learning and inference. I calculated the sufficient statistics for the VM and used them in the updateESS function and use the parameter updates for Mu and K from Bishop's Pattern recognition textbook. I have tested ML and EM for the fully observed case with the Discrete -> Continuous Model and provided a simple toy example. I am in the process of testing EM for the model with missing data.
I have implemented the Von Mises distribution (VM) as another continuous distribution. It is structured in the same way as the Gaussian Distribution in terms of using the same functions for learning and inference. I calculated the sufficient statistics for the VM and used them in the updateESS function and use the parameter updates for Mu and K from Bishop's Pattern recognition textbook. I have tested ML and EM for the fully observed case with the Discrete -> Continuous Model and provided a simple toy example. I am in the process of testing EM for the model with missing data.