Continuous bag of words

Continuous bag of words (CBOW) is a word embedding predicts the centre word from a fixed window size of context words. The input and output to the model are the one-hot code representation of pairs of words. In general:

• It trains faster than skipgram as the task is much simpler.
• It also has better syntactic relationships (i.e., cat and cats).
• Better represents more frequent words.