ReaLX 2018
The SICSA Reasoning, Learning and Explainability Workshop 2018
June 27th 2018
Programme
Invited Talks
Prof Ehud B. Reiter - Aberdeen University
Title: Choosing Words to Communicate Data to People
Abstract: An essential part of explaining reasoning is communicating data. Because the human mind is very good at language (and many people struggle to understand raw numbers and data visualisations), it is often useful to describe data using words. However choosing words to communicate data is not easy, in part because the choice is usually influenced by context; for example the best colour term to describe an RGB value depends on lighting, nearby objects, discourse history, and the preferences of the target user. In this talk I will introduce the problem of choosing words to describe data, give examples to illustrate why it is difficult, and summarise some of the research done in Aberdeen and elsewhere on this topic.
Prof Helen Hastie - Heriot-Watt University
Title: Transparent Robotics through Explainability
Abstract: In the near future, we will see a rapid adoption of robots and autonomous systems in modern society, from self-driving cars to robots in the home and in hazardous and hard-to-reach environments. Trust is an important commodity in this regard because it encourages cooperation with the robot and enables smooth execution of the task at hand, with humans and robots working as part of the same team. There is an important trade-off between control and trust in autonomy and one solution is to increase transparency by having the system explain its reasoning and actions. Prof. Hastie presents techniques for establishing and maintaining transparency through dialogue and explanations, with applications focusing on her work with interfaces for underwater autonomous systems.