Gaussian Mixture Models

• Soft clustering model which gives the probability for a point to belong to a cluster
• A function that comprises of several Gaussians that is defined by
  • A mean $\mu$ for its center
  • A covariance $\Sigma$ for its width
  • A mixing probability $\pi$ for its size
• A condition they follow is that the sum of their mixing probability adds up to 1
• To fit these parameters we use Expectation - Maximization algorithm which finds the solutions to the maximum likelihood #data-science