Measurement test program development problem statement Measuring device gives possible readings and it is assumed that measurement/test is performed with no misinformation, , i.e. . The measurement/test has to be repeated k times, to fully compensate a source entropy by the acquired mutual information. Finding the number of measurement/test repetitions k and then, finding the experiment strategy (described by the decision tree) is the task.
If H ( X ) is the source original entropy (before the first measurement/test) and is the maximum mutual information that can be obtained from single measurement/test, then, the minimum number of measurement/test repetitions, necessary to compensate the source entropy is:
... zobacz całą notatkę
Komentarze użytkowników (0)