Introduction Responding to changing contexts and audiences, people speak and write in different styles. A naturally occurring linguistic variation, style is defined by a large set of attributes of texts which includes but is not limited to formality, politeness, diction, and emotion. Text style transfer (TST) is a task in NLP that aims to change certain attributes in the style of a text while preserving its meaning. A task with a long history, it has significant applications for downstream tasks like machine translation, in which it can help reduce the difficulty of interpreting stylistic texts; text simplification, in which it translates complicated text for a wider audience; and formality transfer, in which help modify the formality of text for different social contexts.
Posts
WARNING: This article was written by the author during high-school, in a non-professional capacity. Meta-learning, or learning to learn, is a paradigm of machine learning algorithms that can generalize itself with meta-knowledge of a certain form such that it can apply to various settings. While it is originally a hallmark of human intelligence, numerous meta-learning perspectives and approaches are springing up in recent years. This paper provides an overview of recent meta-learning approaches, especially for Model-Agnostic Meta-Learning (and its derivatives), Meta-Reinforcement Leaning, and Few-shot (or Zero/One-shot), three emerging methods in the past five years.