Long Text Generation via Adversarial Training with Leaked Information
Explore this paper's citation graph
Summary
The discriminative net is allowed to leak its own high-level extracted features to the generative net to further help the guidance, and LeakGAN is proposed, which is highly effective in long text generation and also improves the performance in short text generation scenarios.
- Type
- preprint
- Published
- 2017-09-24
- Cited by
- 533
- References
- 30
- Access
- Open access
- OpenAlex
- https://openalex.org/W2757836268
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:3389583
Keywords
Computer science, Discriminative model, Closed captioning, Generative grammar, Artificial intelligence
References
- Hierarchical Reinforcement Learning for Adaptive Text Generation
- Text Understanding from Scratch
- Generating Sequences With Recurrent Neural Networks
- Microsoft COCO Captions: Data Collection and Evaluation Server
- Show and tell: A neural image caption generator
- From captions to visual concepts and back
- Long Short-Term Memory
- Dropout: a simple way to prevent neural networks from overfitting
- Bleu: a Method for Automatic Evaluation of Machine Translation
- Between MDPs and Semi-MDPs: A Framework for Temporal Abstraction in Reinforcement Learning
- Chinese Poetry Generation with Recurrent Neural Networks
- Policy Gradient Methods for Reinforcement Learning with Function Approximation
- How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?
- Semantic Rule Based Text Generation
- Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation
- SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient
- Adversarial Learning for Neural Dialogue Generation
- Controllable Text Generation
- Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
- FeUdal Networks for Hierarchical Reinforcement Learning
Cited by
- A Note on the Inception Score
- An Introduction to Image Synthesis with Generative Adversarial Nets
- Neural Text Generation: Past, Present and Beyond
- Generating Diverse and Accurate Visual Captions by Comparative Adversarial Learning
- CoT: Cooperative Training for Generative Modeling
- Toward Diverse Text Generation with Inverse Reinforcement Learning
- Expert-based reward function training: the novel method to train sequence generators
- Deep Reinforcement Learning for Sequence-to-Sequence Models
- Learning Semantic Sentence Embeddings using Sequential Pair-wise Discriminator
- On Accurate Evaluation of GANs for Language Generation
- SentiGAN: Generating Sentimental Texts via Mixture Adversarial Networks
- BFGAN: Backward and Forward Generative Adversarial Networks for Lexically Constrained Sentence Generation
- A Reinforcement Learning Framework for Natural Question Generation using Bi-discriminators
- A Survey of the Usages of Deep Learning for Natural Language Processing
- Abstractive Document Summarisation using Generative Adversarial Networks
- Top-Down Tree Structured Text Generation
- Generating Text through Adversarial Training Using Skip-Thought Vectors
- Generating Classical Chinese Poems via Conditional Variational Autoencoder and Adversarial Training
- Adversarial Text Generation via Feature-Mover's Distance
- Embedding Multimodal Relational Data for Knowledge Base Completion
Related papers
- Improved discriminative training for generative model
- Generative Prompt Model for Weakly Supervised Object Localization
- Adaptive Discriminative Generative Model and Its Applications
- Improving Generative Model-based Unfolding with Schrödinger Bridges
- Generative re-ranking model for dependency parsing of Italian sentences
- SelfEval: Leveraging the discriminative nature of generative models for evaluation
- A discriminative framework for modelling object classes
- Principled Hybrids of Generative and Discriminative Models
- Discriminative Models for Semi-Supervised Natural Language Learning