loading page

Improving Users' Mental Model with Attention-directed Counterfactual Edits
  • +4
  • Kamran Alipour,
  • Arijit Ray,
  • Xiao Lin,
  • Michael Cogswell,
  • Jurgen Schulze,
  • Yi Yao,
  • Giedrius Burachas
Kamran Alipour
University of California San Diego

Corresponding Author:kalipour@eng.ucsd.edu

Author Profile
Arijit Ray
SRI International
Author Profile
Xiao Lin
SRI International
Author Profile
Michael Cogswell
SRI International
Author Profile
Jurgen Schulze
University of California San Diego
Author Profile
Yi Yao
SRI International
Author Profile
Giedrius Burachas
SRI International
Author Profile

Abstract

In the domain of Visual Question Answering (VQA), studies have shown improvement in users’ mental model of the VQA system when they are exposed to examples of how these systems answer certain Image-Question (IQ) pairs. In this work, we show that showing controlled counterfactual image-question examples are more effective at improving the mental model of users as compared to simply showing random examples. We compare a generative approach and a retrieval-based approach to show counterfactual examples. We use recent advances in generative adversarial networks (GANs) to generate counterfactual images by deleting and inpainting certain regions of interest in the image. We then expose users to changes in the VQA system’s answer on those altered images. To select the region of interest for inpainting, we experiment with using both human-annotated attention maps and a fully automatic method that uses the VQA system’s attention values. Finally, we test the user’s mental model by asking them to predict the model’s performance on a test counterfactual image. We note an overall improvement in users’ accuracy to predict answer change when shown counterfactual explanations. While realistic retrieved counterfactuals obviously are the most effective at improving the mental model, we show that a generative approach can also be equally effective.
06 Jun 2021Submitted to Applied AI Letters
18 Jun 2021Submission Checks Completed
18 Jun 2021Assigned to Editor
25 Jun 2021Reviewer(s) Assigned
13 Aug 2021Review(s) Completed, Editorial Evaluation Pending
17 Aug 2021Editorial Decision: Revise Minor
17 Sep 20211st Revision Received
17 Sep 2021Submission Checks Completed
17 Sep 2021Assigned to Editor
04 Oct 2021Reviewer(s) Assigned
12 Oct 2021Review(s) Completed, Editorial Evaluation Pending
25 Oct 2021Editorial Decision: Accept
15 Nov 2021Published in Applied AI Letters. 10.1002/ail2.47