AI Seminar: Towards an AI-Human Symbiosis Using Information Visualization

Speaker

Luana Micallef, Assistant Professor in the Human-Centred Computing Section at Department of Computer Science, University of Copenhagen.

Abstract

Vast amount of data is routinely generated due to advances in technology. Data such as the molecular profile of cancer patients could provide important insights for example drug discovery or precision medicine. Artificial intelligence techniques are helping data scientists reveal patterns in the data. However, a machine cannot replace the human intelligence entirely. The accuracy of the insights revealed by data-driven models can be improved by the human, and the human can learn from the insights derived by the models. Yet such a symbiotic relationship between the machine and the human is only possible through a user interface that allows the user to interaction with the data and an information visualization that is specifically optimized for the human perception.

In this talk, I will demonstrate the importance of information visualization in bridging the two quite distinct fields of data science and human-computer interaction, to help scientists and domain experts (e.g., clinicians) handle, explore and make sense of their data. I will present our latest research in creating visualization techniques that not only communicate the insights predicted by data-driven models, but also allow the human to interact with the data and the models. I will also discuss the importance of optimizing visualization design for human perception.