Sentence examples for the feature evaluation from inspiring English sources

The training and development sets were used to train and refine the machine learning models, and to perform the feature evaluation studies.

However, we can see that the feature evaluation components from the first two groups (CV and BR) tend to dominate the best performing combinations.

As expected from the feature evaluation results, the POS feature is the one with most impact on the results, largely affecting the recall for all entity classes.

Table 1 shows the results of the feature evaluation, using first-order CRF models created with the training set documents and tested on the development set.

Based on the results of the feature evaluation, we used the best feature set (i.e. removing dependency parsing and capitalization features) to train various CRF models of orders 1 to 4. Applied separately, higher-order models produced similar or slightly inferior results when compared to first-order models (between -0.02% and -0.48% reduction in F-measure).

To estimate the importance of each TFBS family on nucleosome position, the feature evaluation algorithm mRMR was used to rank TFBS families according to their relevance to the sample types and redundancy to other features.

Table 2 shows the results obtained on the development set, with the same models used for the overall feature evaluation shown above.

Finally the SEMG feature evaluation index based reported work is a step forward to develop more powerful, flexible, efficient applications leading to prosthetic design.

After the mRMR feature evaluation, a ranking feature set is obtained.

The mRMR feature evaluation uses incremental search methods for optimal features and would loop N rounds when given a feature set with N features.

Listed in Online Supporting Information S5 are two kinds of outputs: the first one is the MaxRel list which shows ranks of features for their relevance to the target; the second is mRMR list showing the mRMR ranks according to the feature order satisfying Eq. 3. In this study, only the mRMR list was used as the results of feature evaluation.