Abstract:With the recent exciting achievements of Google’s artificial intelligence system in the game of Go, deep reinforcement learning (DRL) has witnessed considerable development. DRL combines the abilities of sensing and making decisions provided by deep learning and reinforcement learning. Natural language processing (NLP) involves a large number of vocabularies or statements that have to be represented, and its subtasks, such as the dialogue system and machine translation, involve many decision problems that are difficult to model. Because of the aforementioned reasons, DRL can be appropriately applied to various NLP tasks such as named entity recognition, relation extraction, dialogue system, image caption, and machine translation. Further, DRL is helpful in improving the framework or the training pipeline of the aforementioned tasks, and notable achievements have been obtained. DRL is not an algorithm or a method but a paradigm. Many researchers fit plenty of NLP tasks in this paradigm and achieve better performance. Specifically, in text generation based on the reinforcement learning paradigm, the learning process that is used to produce a predicted sequence from the given source sequence can be considered to be the Markov decision process (MDP). In MDP, an agent interacts with the environment by receiving a sequence of observations and scaled rewards and subsequently produces the next action or word. This causes the text generation model to achieve decision-making ability, which can result in future success. Thus, the text generation task integrated with reinforcement learning is an attractive and promising research field. This study presented a comprehensive introduction and a systemic overview. First, we presented the basic methods in DRL and its variations. Then, we showed the main applications of DRL during the text generation task, trace the development of DRL, and summarized the merits and demerits associated with these applications. The final section enumerated some future research directions of DRL combined with NLP.

Generative AI for Deep Reinforcement Learning: Framework, Analysis, and Use Cases

Deep Reinforcement Learning for NLP.

GAN-Based Interactive Reinforcement Learning from Demonstration and Human Evaluative Feedback

Generative AI for Advanced UAV Networking

Reinforcement Learning for Generative AI: State of the Art, Opportunities and Open Research Challenges

Enhancing Reality: Exploring the Potential of Generative Artificial Intelligence

Research on Knowledge Discovery and Sharing in AIGC Virtual Teaching and Research Room Empowered by Big Data Analysis and Natural Language Processing Algorithms

Robustness and Performance of Deep Reinforcement Learning.

Generative AI for Unmanned Vehicle Swarms: Challenges, Applications and Opportunities

The New Frontier of AI Research: Generative Adversarial Networks

Deep Reinforcement Learning With Application to Air Confrontation Intelligent Decision-Making of Manned/Unmanned Aerial Vehicle Cooperative System

Generative Adversarial Imagination for Sample Efficient Deep Reinforcement Learning

Training and Evaluation of Deep Policies using Reinforcement Learning and Generative Models

Generative AI for Self-Adaptive Systems: State of the Art and Research Roadmap

Next-Generation Wi-Fi Networks with Generative AI: Design and Insights

Research Progress of Deep Reinforcement Learning Applied to Text Generation

Deep Reinforcement Learning: Framework, Applications, and Embedded Implementations

Generative AI: A systematic review using topic modelling techniques

Collaboration with Generative Artificial Intelligence: An Exploratory Study Based on Learning Analytics

Generative Artificial Intelligence and GPT using Deep Learning: A Comprehensive Vision, Applications Trends and Challenges