Show Me What You’re Looking For
Leonid Schwenke, Martin Atzmueller · Proceedings of the ... International Florida Artificial Intelligence Research Society Conference · 2021
While Transformers have shown their advantages consideringtheir learning performance, their lack of explainabilityand interpretability is still a major problem.This specifically relates to the processing of time series,as a specific form of complex data. In this paper,we propose an approach for visualizing abstracted informationin order to enable computational sensemakingand local interpretability on the respective Transformermodel. Our results demonstrate the efficacy ofthe proposed abstraction method and visualization, utilizingboth synthetic and real world data for evaluation.