Finite Gaussian Mixture

For each class, we draw and remember a mean and variance. For each object, we draw (and remember) one of the classes, then draw from a Gaussian with the corresponding mean and variance.

(define (cls-distribution)
  (uniform-draw (list 'cls-a 'cls-b 'cls-c)))

(define object->cls
  (mem (lambda (object) (cls-distribution))))

(define cls->gaussian-parameters
  (mem (lambda (cls) (list (gaussian 65 10) (uniform 0 8)))))

(define (observe object)
  (apply gaussian (cls->gaussian-parameters (object->cls object))))

(map observe '(tom dick harry bill fred))