Humpty Dumpty: Controlling Word Meanings via Corpus Poisoning
IEEE Symposium on Security and Privacy, pp. 1295-1313, 2020.
Word embeddings, i.e., low-dimensional vector representations such as GloVe and SGNS, encode word \"meaning\" in the sense that distances between words’ vectors correspond to their semantic proximity. This enables transfer learning of semantics for a variety of natural language processing tasks.Word embeddings are typically trained on lar...More
PPT (Upload PPT)