EL-VIT: Probing vision transformer with interactive visualization

• Main Authors: ZHOU, Hong, ZHANG, Rui, LAI, Peifeng, GUO, Chaoran, WANG, Yong, SUN, Zhida, LI, Junjie
• 格式: text
• 語言: English
• 出版: Institutional Knowledge at Singapore Management University 2023
• 主題: Education Tool; Explainable AI; Vision Transformer; Visual Analysis; Databases and Information Systems; Numerical Analysis and Scientific Computing
• 在線閱讀: https://ink.library.smu.edu.sg/sis_research/8708; https://ink.library.smu.edu.sg/context/sis_research/article/9711/viewcontent/EL_VIT_av.pdf
• 機構: Singapore Management University

Nowadays, Vision Transformer (ViT) is widely utilized in various computer vision tasks, owing to its unique self-attention mechanism. However, the model architecture of ViT is complex and often challenging to comprehend, leading to a steep learning curve. ViT developers and users frequently encounter difficulties in interpreting its inner workings. Therefore, a visualization system is needed to assist ViT users in understanding its functionality. This paper introduces EL-VIT, an interactive visual analytics system designed to probe the Vision Transformer and facilitate a better understanding of its operations. The system consists of four layers of visualization views. The first three layers include model overview, knowledge background graph, and model detail view. These three layers elucidate the operation process of ViT from three perspectives: the overall model architecture, detailed explanation, and mathematical operations, enabling users to understand the underlying principles and the transition process between layers. The fourth interpretation view helps ViT users and experts gain a deeper understanding by calculating the cosine similarity between patches. Our two usage scenarios demonstrate the effectiveness and usability of EL-VIT in helping ViT users understand the working mechanism of ViT.