| | NoDC/NoMut | NoDC/Mut | DC/NoMut | DC/Mut |
|---|
Recombination | SENS. | PPV | SENS. | PPV | SENS. | PPV | SENS. | PPV |
Link R. (High Rate) | 43 ± 4 | 61 ± 12 | 61 ± 6 | 58 ± 8 | 63 ± 3 | 84 ± 5 | 68 ± 4 | 74 ± 8 |
Link R. (Low Rate) | 18 ± 7 | 18 ± 8 | 42 ± 8 | 22 ± 8 | 68 ± 4 | 82 ± 9 | 68 ± 4 | 80 ± 6 |
Parental R. (High Rate) | 23 ± 7 | 26 ± 8 | 56 ± 7 | 38 ± 4 | 48 ± 3 | 68 ± 10 | 66 ± 4 | 69 ± 6 |
Parental R. (Low Rate) | 12 ± 5 | 12 ± 5 | 33 ± 6 | 14 ± 4 | 61 ± 4 | 79 ± 6 | 60 ± 2 | 63 ± 7 |
- We report the ability of link and parental recombination to explore the search space for high (0.4) and low (0.1) exchange rates. Results for link-recombination with high and low exchange rates are respectively presented in rows 1 and 2. Results for parental-recombination with high and low exchange rates are respectively shown in rows 3 and 4. In the first column (NoDC/NoMut), we consider the results for these various recombination strategies when neither mutation nor Deterministic Crowding were used. We only use mutation or Deterministic Crowding in column 2 (NoDC/Mut) and column 3 (DC/NoMut), respectively. Mutation and Deterministic Crowding are used jointly in column 4 (DC/Mut). In a given cell, the figures correspond to the results obtained across 10 runs performed on different and independent datasets with a size of 300. These results are presented as the mean ± the standard deviation of the sensitivity (SENS.) and positive predictive value (PPV) across the ten runs. For the sake of clarity, they are expressed as percentages.
