HackerPapers

arXiv:1606.02245cs.CL

Iterative Alternating Neural Attention for Machine Reading

Alessandro Sordoni, Philip Bachman, Adam Trischler, Yoshua Bengio

Abstract

We propose a novel neural attention architecture to tackle machine comprehension tasks, such as answering Cloze-style queries with respect to a document. Unlike previous models, we do not collapse the query into a single vector, instead we deploy an iterative alternating attention mechanism that allows a fine-grained exploration of both the query and the document. Our model outperforms state-of-the-art baselines in standard machine comprehension benchmarks such as CNN news articles and the Children's Book Test (CBT) dataset.

arXiv comments
Journal reference
DOI
Subjects
cs.CLComputation and Language cs.NENeural and Evolutionary Computing

On Hacker News

Linked by 2 identifiable commenters in 2 comments across 1 thread, from Jun 2016 to Jun 2016.

Comments linking this paper, per year.

Discussed in comments1 thread

  1. Highest results yet for machine reading comprehension on two benchmark tests 2 comments

    • Interestingly, another research team published slightly better results across the board on the same data sets, on the same day [1]. Also, both teams are Canadian, I believe. [1] http://arxiv.org/abs/1606.02245

      svantanaView on HNParent context

    • You are correct - both teams are Canadian and part of Maluuba Research Lab in Montreal. [www.maluuba.com] Here is the difference between the two: The EpiReader model uses a two-stage process to arrive at an answer. The first stage extracts a small set of candidate answers from…

      rahulmehrotraView on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.