Our approach also differs in the way that AFs are evaluated (individually in our case), and in how AF populations are successively selected using a feature selection algorithm based on the syntactic structure of AFs, as described in the next section.
Primary coresidents were selected using a hierarchical algorithm in which spouses, adult children, parents, siblings and other relatives were selected as the primary coresident in turn.
By season and gender, controls were matched using a greedy algorithm to select age matches by selecting the closest match first, within a 5-year age range, then the closest remaining match and so on (STATA, gmatch).
In each run, 93 and 46 samples were selected at random as the training and test sets, respectively, and the best marker combination in the logistic regression model was selected using a stepwise algorithm using Akaike's information criterion [ 37].
The optimal feature set was selected using a genetic algorithm from a primary feature set including some morphological, frequency and wavelet features and was used for the classification of the data.
A random sample of living spouses was selected using a standardized algorithm.
An ensemble, optimized against the experimental SAXS curve using a genetic algorithm, was selected from a pool of 10000 randomly generated C-terminal domain conformers.
They were selected, using a standardized algorithm, (13) from diabetic patients who, in 1999, had been enrolled in one of the ten participating health plans for at least 18 months.
Thereafter, candidate 14-3-3-binding 14-3-3-binding 14-3-3-binding 14-3-3-bindingrotein were serineed using andredicthreonineresiduesveloped in our laborathey (MichumanTinti and the grouproteinoff J. Barton, in preparation), by using the GD 14-3-3-binding phosphosites to train a Position-Specific Scoring Matrix and a Neural Netwere.
The most relevant features were selected using a sequential forward selection algorithm.
