ReasoNet: Learning to Stop Reading in Machine Comprehension

By Yelong Shen, Po-Sen Huang, Jianfeng Gao, Weizhu Chen. 2016

309 citations recorded by the source corpus.

Teaching a computer to read and answer general questions pertaining to a document is a challenging yet unsolved problem. In this paper, we describe a novel neural network architecture called the Reasoning Network (ReasoNet) for machine comprehension tasks. ReasoNets make use of multiple turns to effectively exploit and then reason over the relation among queries, documents, and answers. Different from previous approaches using a fixed number of turns during inference, ReasoNets introduce a termination state to relax this constraint on the reasoning depth. With the use of reinforcement learning, ReasoNets can dynamically determine whether to continue the comprehension process after digesting intermediate results, or to terminate reading when it concludes that existing information is adequate to produce an answer. ReasoNets achieve superior performance in machine comprehension datasets, including unstructured CNN and Daily Mail datasets, the Stanford SQuAD dataset, and a structured Graph Reachability dataset.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, Question Answering & Reading Comprehension

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

In Question Answering & Reading Comprehension

Citations in the corpus

Multi-Granularity Hierarchical Attention Fusion Networks for Reading Comprehension and Question Answering (2018, background): “, 2017b) and (Shen et al., 2017), the multi-turn memory networks are designed to simulate multi-step reasoning in machine reading comprehension.”
U-Net: Machine Reading Comprehension with Unanswerable Questions (2018, background): “…Currently, end-to-end neural network models have achieved great successes for machine reading comprehension (Seo et al. 2016; Kumar et al. 2015; Sukhbaatar et al. 2015; Cui et al. 2016; Xiong, Zhong, and Socher 2016; Dhingra et al. 2016; Shen et al. 2016; Hu et al. 2017; Wang, Yan, and Wu 2018).”
Bidirectional Attention Flow for Machine Comprehension (2016, methodology): “Unlike previously popular attention mechanisms (Weston et al., 2015; Hill et al., 2016; Sordoni et al., 2016; Shen et al., 2016), the attention ﬂow layer is not used to summarize the query and context into single feature vectors.”
MS MARCO: A Human Generated MAchine Reading COmprehension Dataset (2016, background): “• ReasoNet: ReasoNet [28] also relies on attention, but is also a dynamic multi-turn model that attempts to exploit and reason over the relation among queries, contexts and answers.”
Real-Time Open-Domain Question Answering with Dense-Sparse Phrase Index (2019, background): “popularity (Shen et al., 2017; Raiman and Miller, 2017; Min et al., 2018; Das et al., 2019), where a few (5-10) documents relevant to the question are retrieved and then a deep neural model finds the answer in the document.”
Multi-step Retriever-Reader Interaction for Scalable Open-domain Question Answering (2019, background): “Models that do iterative reasoning (Shen et al., 2017; Liu et al., 2017) i.”
Denoising Distantly Supervised Open-Domain Question Answering (2018, background): “Many reading comprehension systems (Chen et al., 2016; Dhin-gra et al., 2017a; Cui et al., 2017; Shen et al., 2017; Wang et al., 2017) have been proposed and achieved promising results since their multi-layer architectures and attention mechanisms allow them to reason for the question.”
The Effect of Natural Distribution Shift on Question Answering Models (2020, neutral): “ReasoNet (Single) (Shen et al., 2017) https://worksheets.”
Adversarial Examples for Evaluating Reading Comprehension Systems (2017, neutral): “We also validate our major findings on twelve other published models with publicly available test-time code: ReasoNet Single and Ensemble versions (Shen et al., 2017), Mnemonic Reader Single and Ensemble versions (Hu et al., 2017), Structural Embedding of Dependency Trees (SEDT) Single and Ensemble…”
Dynamic Neural Networks: A Survey (2021, background): “Language Sa & Te Language modeling [11], [16], [118], [170], [172], Machine translation [16], [35], [36], Classification [64], [65], [174], Processing Sentiment analysis [168], [169], [171], [175], [176], Question answering [35], [63], [168], [171], [173]”
Gated Self-Matching Networks for Reading Comprehension and Question Answering (2017, background): “End-to-end Neural Networks for Reading Comprehension Along with cloze-style datasets, several powerful deep learning models (Hermann et al., 2015; Hill et al., 2016; Chen et al., 2016; Kadlec et al., 2016; Sordoni et al., 2016; Cui et al., 2016; Trischler et al., 2016; Dhingra et al., 2016; Shen et ”
Constructing Datasets for Multi-hop Reading Comprehension Across Documents (2017, background): “…thereof (Peng et al., 2015; Weissenborn, 2016; Xiong et al., 2016; Liu and Perez, 2017), which often is iterated over several times (Sordoni et al., 2016; Mark et al., 2016; Seo et al., 2016; Hu et al., 2017) and may contain train-able stopping mechanisms (Graves, 2016; Shen et al., 2017b).”
Neural Approaches to Conversational AI (2018, neutral): “easoning. Multi-step reasoning models are pioneered by Hill et al. (2015); Dhingra et al. (2016); Sordoni et al. (2016); Kumar et al. (2016), who used a pre-determined ﬁxed number of reasoning steps. Shen et al. (2017b,c) showed that multi-step reasoning outperforms single-step ones and dynamic mult”
Gated-Attention Readers for Text Comprehension (2016, background): “The success of many recent models can be attributed primarily to two factors: (1) Multi-hop architectures (Weston et al., 2015; Sordoni et al., 2016; Shen et al., 2016), allow a model to scan the document and the question iteratively for multiple passes.”
Stochastic Answer Networks for Machine Reading Comprehension (2017, methodology): “Later, Shen et al. (2016) showed that dynamically determining the number of steps based on the complexity of the question and answer leads to improvements in MRC; the model uses reinforcement learning to predict the number of reasoning steps for each question-answer pair.”
Recent Trends in Deep Learning Based Natural Language Processing (2017, methodology): “mbedding 0.73 Weston et al. [112] Memory networks 93.3 0.83 Sukhbaatar et al. [138] End-to-end memory networks 88.4 Kumar et al. [102] DMN 93.6 Chen et al. [168] Document Reader 70.0/79.0 Shen et al. [169] ReasoNet 69.1/78.9 Yu et al. [170] QAnet 76.2/84.6 H. Dialogue Systems Two types of dialogue s”
Efficient and Robust Question Answering from Minimal Context over Documents (2018, background): “, 2016), ReasoNet (Shen et al., 2017) and Mnemonic Reader (Hu et al.”