Two-Stage Synthesis Networks for Transfer Learning in Machine Comprehension
arXiv:1706.09789
Abstract
We develop a technique for transfer learning in machine comprehension (MC) using a novel two-stage synthesis network (SynNet). Given a high-performing MC model in one domain, our technique aims to answer questions about documents in another domain, where we use no labeled data of question-answer pairs. Using the proposed SynNet with a pretrained model from the SQuAD dataset on the challenging NewsQA dataset, we achieve an F1 measure of 44.3% with a single model and 46.6% with an ensemble, approaching performance of in-domain models (F1 measure of 50.0%) and outperforming the out-of-domain baseline of 7.6%, without use of provided annotations.
References in corpus (16)
- Adam: A Method for Stochastic Optimization
- SQuAD: 100,000+ Questions for Machine Comprehension of Text
- The Goldilocks Principle: Reading Children's Books with Explicit Memory Representations
- Dynamic Coattention Networks For Question Answering
- Machine Comprehension Using Match-LSTM and Answer Pointer
- Multimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding
- Simple Baseline for Visual Question Answering
- Learning Recurrent Span Representations for Extractive Question Answering
- A Thorough Examination of the CNN/Daily Mail Reading Comprehension Task
- Multi-Perspective Context Matching for Machine Comprehension
- Latent Predictor Networks for Code Generation
- Character-Level Question Answering with Attention
- NewsQA: A Machine Comprehension Dataset
- Knowing When to Look: Adaptive Attention via A Visual Sentinel for Image Captioning
- Semi-Supervised QA with Generative Domain-Adaptive Nets
- Data Recombination for Neural Semantic Parsing