body or inappropriate contact of the sensors, namely artifact.
At this first stage the output is raw signal. In the second stage,
i.e. feature extraction stage, the raw signal obtained from the
previous stage is converted into a feature vector. The feature
vector represents relevant structure in the raw data. Then, a
process called dimensionality reduction is carried out, in
which redundant information is elimated from the feature
vector, generating a reduced feature vector [3]. The third
stage is classification, i.e. translation algorithm, in which
categories are identified from the reduced feature vector by
employing pattern recognition techniques. Finally, in the
fourth stage, i.e. the controller, the categories obtained from
the classification stage are translated into control commands
for execution