Evaluator for symbolic regression models for learning a variance model

[gkronber]

r14528: added an evaluator for symbolic regression models which calculates the likelihood that variable values are sampled from a zero mean Gaussian distribution where the variance is given by the model. This can be used to learn input-dependent variances.

[gkronber]

Linear scaling needs to be turned off.

[gkronber]

The objective is to maximize the following log likelihood that the observed residuals stem from a zero-mean normal distribution with std. dev. s = f(x).

      var l2pi = Math.Log(2.0 * Math.PI);
      var ll = -0.5 *
               boundedEstimatedValues.Zip(targetValues, (s, t) =>
                 +l2pi
                 + Math.Log(s * s)
                 + (t * t) / (s * s)
                 ).Sum();

      return ll;

This is problematic as the sign of s has no effect on the result (same for the sign of t). It would probably be better to model the variance s² directly. However, in this case negative values for s² are not possible and must be prevented. On option would be to produce a model for log(s²) instead.

[gkronber]

Should either be released or removed from trunk

[gkronber]

r15198: removed evaluator for learning of variance models from the trunk to prepare for release

[gkronber]

r15199: added merge info for r14528 and r15198 to stable to align with trunk