Using multiple samples to learn mixture models

In the mixture models problem it is assumed that there are K distributions θ1,…,θK and one gets to observe a sample from a mixture of these distributions with unknown coefficients. The goal is to asso