1. Under the Bayesian decision rule, the classification error is given by P(error) = p(errorl)p(x)dx = min[P(W1lx),P(w2l] p(x)dx Show that for arbitrary density functions, an upper bound of the classification error can be found by replacing min [P(W1|x), P(wz|x)] with 2P(W1|x)P(wz|x). And a lower bound can be found by replacing min [P(W1|x), P(wz|x)] with P(W1|x)P(W2|x)