WIP: Compute the norm minDCF and fix the floating-point precision

Change the default floating-point precision for minDCF

@ssarfjoo this is not intended fix. As discussed before, we have an option for this: --decimal to change the precision. Instead of changing the hard coded value from 1 to 4, it's better to use that option value instead.

The default value must be changed and the implementation for computing the minDCF is not optimized. Without setting the --decimal it must report with a four-digit floating point.

Then it may be a good idea to use --decimal + 3 (which defaults to 1) for mindcf.

Is this possible to set the default --decimal for each criterion separately?

The way it is implemented currently, I don't think so.

assigned to @ssarfjoo

marked as a Work In Progress

closed