The Transform-and-Perform framework: Explainable deep learning beyond classification

Vidya Prasad, Ruud J. G. van Sloun, Stef van den Elzen, Anna Vilanova, Nicola Pezzotti

DOI: 10.1109/TVCG.2022.3219248

Room: 109

2023-10-25T04:57:00ZGMT-0600Change your timezone on the schedule page
2023-10-25T04:57:00Z
Exemplar figure, described by caption below
We introduce the Transform-and-Perform (T&P) framework, designed to assist visual analytics (VA) designers in creating VA systems with general applicability to high-dimensional-to-high-dimensional (H-H) problems. T&P helps identify workflows and analysis strategies for designing new VA systems. It also helps reveal potential gaps in existing systems. T&P enables analysis across the "3Ws" of model behavior: 1) when a behavior occurs (input analysis), 2) how & why it occurs (model analysis), and 3) what the behavior is (output analysis). By utilizing input-output symmetries, T&P offers a formal approach to understanding a model's inductive biases, crucial for analyzing a range of DL models.
Fast forward
Full Video
Keywords

Visual Analytics;Explainable AI;XAI;Framework;Deep Learning;High-dimensional-to-high-dimensional translation

Abstract

In recent years, visual analytics (VA) has shown promise in alleviating the challenges of interpreting black-box deep learning (DL) models. While the focus of VA for explainable DL has been mainly on classification problems, DL is gaining popularity in high-dimensional-to-high-dimensional (H-H) problems such as image-to-image translation. In contrast to classification, H-H problems have no explicit instance groups or classes to study. Each output is continuous, high-dimensional, and changes in an unknown non-linear manner with changes in the input. These unknown relations between the input, model and output necessitate the user to analyze them in conjunction, leveraging symmetries between them. Since classification tasks do not exhibit some of these challenges, most existing VA systems and frameworks allow limited control of the components required to analyze models beyond classification. Hence, we identify the need for and present a unified conceptual framework, the Transform-and-Perform framework (T&P), to facilitate the design of VA systems for DL model analysis focusing on H-H problems. T&P provides a checklist to structure and identify workflows and analysis strategies to design new VA systems, and understand existing ones to uncover potential gaps for improvements. The goal is to aid the creation of effective VA systems that support the structuring of model understanding and identifying actionable insights for model improvements. We highlight the growing need for new frameworks like T&P with a real-world image-to image translation application. We illustrate how T&P effectively supports the understanding and identification of potential gaps in existing VA systems.