PAC-Bayes Analysis of Sentence Representation
This work provides a theoretical foundation for sentence representation learning, which is incremental but addresses a gap in understanding for NLP practitioners.
The paper tackles the problem of theoretically understanding simple heuristics for learning sentence vectors from unlabeled data, showing that methods like averaging and IDF-weighted averaging are derived from a PAC-Bayes analysis and proposing new algorithms based on this framework.
Learning sentence vectors from an unlabeled corpus has attracted attention because such vectors can represent sentences in a lower dimensional and continuous space. Simple heuristics using pre-trained word vectors are widely applied to machine learning tasks. However, they are not well understood from a theoretical perspective. We analyze learning sentence vectors from a transfer learning perspective by using a PAC-Bayes bound that enables us to understand existing heuristics. We show that simple heuristics such as averaging and inverse document frequency weighted averaging are derived by our formulation. Moreover, we propose novel sentence vector learning algorithms on the basis of our PAC-Bayes analysis.