Inferring coin biases in a model with a latent selector variable.
The generative model: 0. Create coin "A" and coin "B"
- Pick a coin with some probability
- Toss the selected coin "n" times and record the number of heads.
- Repeat the experiment "m" times
- Return the m-vector of head counts
The program infers coin "A" and coin "B" biases given that the selector is unobserved. The coin picking probability is inferred as an auxilliary task.
Updated to use the open source Infer.NET version.