Choosing the parameters of a classifier (and perhaps also the family of class ifiers) from the training set.
  Least False Parameter
value. If the parametric family does not contain the true distribution. this is the best possible parameter value in the sense of minimum Kullback-Leibler divergence
The probability density of the observations, viewed as a function of the parameters, not of the observations.
The logistic distribution has cumulative distribution function and this unction si called the logistic function. Logistic regression and discrimination are based on converting predictions to probabilities through the logistic function.
An L space is the space of random variables X such that or a space of functions f for which in L means
Learning Vector Quantization. A method of designing examples for use in nearest-neighbour procedures.