【论文整理】DL paper notes 2015-11

2015-11

NLP

• Deep Reinforcement Learning with a Natural Language Action Space [arXiv]
• Sequence Level Training with Recurrent Neural Networks [arXiv]
• Teaching Machines to Read and Comprehend [arxiv]
• Semi-supervised Sequence Learning [arXiv]
• Multi-task Sequence to Sequence Learning [arXiv]
• Alternative structures for character-level RNNs [arXiv]
• Larger-Context Language Modeling [arXiv]
• A Unified Tagging Solution: Bidirectional LSTM Recurrent Neural Network with Word Embedding [arXiv]
• Towards Universal Paraphrastic Sentence Embeddings [arXiv]
• BlackOut: Speeding up Recurrent Neural Network Language Models With Very Large Vocabularies [arXiv]
• Sequence Level Training with Recurrent Neural Networks [arXiv]
• Natural Language Understanding with Distributed Representation [arXiv]
• sense2vec - A Fast and Accurate Method for Word Sense Disambiguation In Neural Word Embeddings [arXiv]
• LSTM-based Deep Learning Models for non-factoid answer selection [arXiv]

Programs

• Neural Random-Access Machines [arxiv]
• Neural Programmer: Inducing Latent Programs with Gradient Descent [arXiv]
• Neural Programmer-Interpreters [arXiv]
• Learning Simple Algorithms from Examples [arXiv]
• Neural GPUs Learn Algorithms [arXiv] [code]
• On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models [arXiv]

Vision

• ReSeg: A Recurrent Neural Network for Object Segmentation [arXiv]
• Deconstructing the Ladder Network Architecture [arXiv]
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [arXiv]
• Multi-Scale Context Aggregation by Dilated Convolutions [arXiv] [code]

General

• Towards Principled Unsupervised Learning [arXiv]
• Dynamic Capacity Networks [arXiv]
• Generating Sentences from a `ous Space [arXiv]
• Net2Net: Accelerating Learning via Knowledge Transfer [arXiv]
• A Roadmap towards Machine Intelligence [arXiv]
• Session-based Recommendations with Recurrent Neural Networks [arXiv]
• Regularizing RNNs by Stabilizing Activations [arXiv]