AI Chat AI Image Generator AI Video Text to Speech

Does it care what you asked? Understanding Importance of Verbs in Deep Learning QA System

09/11/2018

∙

by Barbara Rychalska, et al.

∙

∙

In this paper we present the results of an investigation of the importance of verbs in a deep learning QA system trained on SQuAD dataset. We show that main verbs in questions carry little influence on the decisions made by the system - in over 90 change system decision. We track this phenomenon down to the insides of the net, analyzing the mechanism of self-attention and values contained in hidden layers of RNN. Finally, we recognize the characteristics of the SQuAD dataset as the source of the problem. Our work refers to the recently popular topic of adversarial examples in NLP, combined with investigating deep net structure.

Barbara Rychalska
13 publications
Dominika Basaj
7 publications
Przemyslaw Biecek
54 publications
Anna Wroblewska
22 publications

research

∙ 03/29/2019

A General FOFE-net Framework for Simple and Effective Question Answering over Knowledge Bases

Question answering over knowledge base (KB-QA) has recently become a pop...

4 Dekun Wu, et al. ∙

research

∙ 05/26/2019

Gated Group Self-Attention for Answer Selection

Answer selection (answer ranking) is one of the key steps in many kinds ...

0 Dong Xu, et al. ∙

research

∙ 05/02/2020

DeFormer: Decomposing Pre-trained Transformers for Faster Question Answering

Transformer-based QA models use input-wide self-attention – i.e. across ...

0 Qingqing Cao, et al. ∙

research

∙ 04/09/2022

Extending the Scope of Out-of-Domain: Examining QA models in multiple subdomains

Past works that investigate out-of-domain performance of QA systems have...

0 Chenyang Lyu, et al. ∙

research

∙ 10/19/2020

Optimism in the Face of Adversity: Understanding and Improving Deep Learning through Adversarial Robustness

Driven by massive amounts of data and important advances in computationa...

5 Guillermo Ortiz-Jiménez, et al. ∙

research

∙ 02/18/2022

Deep-Learning Architectures for Multi-Pitch Estimation: Towards Reliable Evaluation

Extracting pitch information from music recordings is a challenging but ...

0 Christof Weiß, et al. ∙

research

∙ 09/19/2023

What Learned Representations and Influence Functions Can Tell Us About Adversarial Examples

Adversarial examples, deliberately crafted using small perturbations to ...

0 Shakila Mahjabin Tonni, et al. ∙