Predictive Visual Analytics - Explainability and Trust
Problem
The term predictive visual analytics is used for visual analytics support for predictive analytics, where predictive analytics is used as an umbrella term for prediction techniques from statistical modeling, machine learning, and data mining. In a survey article on predictive visual analytics, Lu et al. [LGH + 17] give an overview on the visual analytics systems and techniques applied for predictive analytics found in the literature, they also offered an outlook for future challenges and research directions in predictive visual analytics. Lu et al. considered deep learning approaches for prediction as a current trend. This trend of such increasingly large and complex models increases the challenges in interpretability and trust in such models. The buzzword (XAI) explainable AI (artificial intelligence) has been used for research efforts in trying to deal with the challenge of interpretability and ability to explain in this regard.
Aim
You should provide a systematic overview of existing approaches in scientific literature that employ visual analytics for increasing explainability and trust in predictive visual analytics solutions.
Other information
References:
[LGH + 17] Y. Lu, R. Garcia, B. Hansen, M. Gleicher, and R. Maciejewski, “The State-of-the-Art in Predictive Visual Analytics,” Computer Graphics Forum, vol. 36, no. 3, pp. 539–562, 2017, doi: 10.1111/cgf.13210.