*3.2. Classifier*

The classifier has the ability to learn the concept during the training phase, mine the situations and assign the context label during the recognition phase in real time. The extracted feature vectors (i.e., Section 3.1.2) are provided to the classifier for the classification. In the first stage, accelerometer data are classified into "active" or "still". In the second stage, audio data are classified by the classifier. In the audio data classifier, we take into consideration only two contexts, while the rest of the contexts are recognized as "sedentary-unknown context". The following sections explain the training and testing of the classifier.
