Part of Advances in Neural Information Processing Systems 8 (NIPS 1995)
Stephen Omohundro
"Family discovery" is the task of learning the dimension and struc(cid:173) ture of a parameterized family of stochastic models. It is espe(cid:173) cially appropriate when the training examples are partitioned into "episodes" of samples drawn from a single parameter value. We present three family discovery algorithms based on surface learn(cid:173) ing and show that they significantly improve performance over two alternatives on a parameterized classification task.