During supervised training, the data is manually transcribed, i.e., the original transcription is exploited during training.
Generally, two methods to train a system exist: supervised learning where data is manually tagged as 'normal' or 'anomalous' and unsupervised learning where the detection system itself has to evaluate which data is normal and which is not.
2) By comparing different data transformation techniques (Additional file 6a), supervised data discretization was shown to be substantially better for improving classifier performance than other methods.
In semi-supervised approaches, the transcribed data is augmented with additional untranscribed data and a supervised training signal is generated by labeling the untranscribed data with 0/1 labels based on the class having the maximum prediction output by the current network.
This type of learning can either be supervised (classification), where the data is labeled to determine how powerful the algorithm is at learning the solution to the problem, or unsupervised (clustering), where the data is unlabeled and the system forms natural groupings (clusters) of patterns automatically (Duda et al. 2000).
That is accomplished almost entirely in a supervised fashion, where labeled data is training an appropriately built neural network.
There are two basic approaches – supervised learning, where the data is pre-classified according to some hypothesis and unsupervised learning where the data is unclassified (usually, but not always, because the potential classes are a priori unknown).
The highlighted category (alignment-free supervised classification of metagenome data) is that of the current work.
The aim of supervised classification of microarray data is to detect genes that might prospectively predict defined outcomes.
During semi-supervised training, the auxiliary data is un-transcribed, i.e., the data is exploited in a semi-supervised fashion.
