Following great success in the image processing field, the idea of
adversarial training has been applied to tasks in the natural language
processing (NLP) field. One promising approach directly applies adversarial
training developed in the image processing field to the input word embed