Neural Networks for Classifying Race
Pro:
Recall that, given reasonable assumptions, two-layer neural networks are capable of approximating arbitrary decision boundaries. Theoretically then, such networks should be capable of yielding good performance on the race classification task.
Recall that, given reasonable assumptions, two-layer neural networks are capable of approximating arbitrary decision boundaries.
Theoretically then, such networks should be capable of yielding good performance on the race classification task.
Con:
Neural networks require a comparatively vast amount of training data. Many (many) free parameters require estimation, especially with large numbers of hidden units.
Neural networks require a comparatively vast amount of training data.
Many (many) free parameters require estimation, especially with large numbers of hidden units.
