PRML classifier: Posterior probability under each Poisson mixtures hypotheses.

prml_classifier(xs_bn, xs_a, xs_b, mu_l = "min", mu_u = "max", e = 0,
  r_a = 0.5, s_a = 2e-10, r_b = 0.5, s_b = 2e-10, n_gq = 20,
  n_per = 100)

Arguments

xs_bn	A vector. Spike counts of repeated dual-stimuli trial data AB.
xs_a	A vector. Spike counts of repeated single-stimulus trial data A.
xs_b	A vector. Spike counts of repeated single-stimulus trial data B.
mu_l	A number. Lower bound of spike counts. "min" by default. Indicating \( max(0, min_{j=A,B,AB}(min(Y_j)-2{std}(Y_j))) \)
mu_u	A number. Upper bound of spike counts. "max" by default. Indicating \( {max_{j=A,B,AB}}(max(Y_j)+2{std}(Y_j)) \)
e	A number. 0 by default. Shringkage on the domain and meansurement of mixing density f under the Intermediate and Mixture hypothese.
r_a	A number. The parameter in gamma prior of spike rate mu_A. rate. Jeffereys' prior by default.
s_a	A number. The parameter in gamma prior of spike rate mu_A. shape. Jeffereys' prior by default.
r_b	A number. The parameter in gamma prior of spike rate mu_B. rate. Jeffereys' prior by default.
s_b	A number. The parameter in gamma prior of spike rate mu_B. shape. Jeffereys' prior by default.
n_gq	A number. 20 by default. Number of grids in Gaussion quadrature.
n_per	A number. 100 by default. Permutation of likihood estimation to obtain the order-invariant estimator.

A list.

post.prob: posterior probabilities under Mixture, Intermediate, Outside, Single hypotheses.
win.model: the model has largest post.prob.