• Publications
  • Influence
Attention is not Explanation
TLDR
We empirically investigate the relationship between attention weights, inputs, and outputs in neural NLP models. Expand
  • 305
  • 23
  • PDF
ERASER: A Benchmark to Evaluate Rationalized NLP Models
TLDR
We propose the Evaluating Rationales And Simple English Reasoning (ERASER) benchmark to advance research on interpretable models in NLP that reveal the `reasoning' behind model outputs. Expand
  • 67
  • 18
  • PDF
Structured Disentangled Representations
TLDR
We propose a two-level hierarchical objective to control relative degree of statistical independence between blocks of variables and individual variables within blocks. Expand
  • 76
  • 7
  • PDF
Learning to Faithfully Rationalize by Construction
TLDR
In many settings it is important for one to be able to understand why a model made a particular prediction. Expand
  • 20
  • 4
  • PDF
Hierarchical Disentangled Representations
TLDR
We introduce a generalization of the evidence lower bound that explicitly represents the trade-offs between sparsity of the latent code, bijectivity of representations, and coverage of the support of the empirical data distribution. Expand
  • 18
  • 3
An Analysis of Attention over Clinical Notes for Predictive Tasks
TLDR
We show, using two EMR corpora and four different predictive tasks, that: (i) inclusion of attention mechanisms is critical for neural encoder modules that operate over notes fields in order to yield competitive performance, but, (ii) it is decidedly less clear whether they provide meaningful support for predictions. Expand
  • 7
  • 2
  • PDF
Question Answering over Knowledge Base using Factual Memory Networks
TLDR
We introduce Factual Memory Network, which learns to answer questions by extracting and reasoning over relevant facts from a Knowledge Base using multi-hop reasoning and refinement. Expand
  • 44
  • 1
  • PDF
Cross Lingual Sentiment Analysis using Modified BRAE
TLDR
We use the Recursive Autoencoder architecture to develop Cross Lingual Sentiment Analysis (CLSA) tool using sentence aligned corpora between a pair of resource rich (English) and resource poor (Hindi) language. Expand
  • 18
  • 1
  • PDF
SciREX: A Challenge Dataset for Document-Level Information Extraction
TLDR
We introduce SciREX, a document level IE dataset that encompasses multiple IE tasks, including salient entity identification and document level $N$-ary relation identification from scientific articles. Expand
  • 13
  • 1
  • PDF
Learning Disentangled Representations of Texts with Application to Biomedical Abstracts
TLDR
We propose a method for learning disentangled representations of texts that code for distinct and complementary aspects, with the aim of affording efficient model transfer and interpretability. Expand
  • 15
  • PDF
...
1
2
...