Aims & Scope It is widely recognised that
the use by people of intelligent tools is one of the key drivers of
technological development. This has been termed Augmented
Intelligence and is a critical element in second generation
Artificial Intelligence (AI), sometimes called AI2.0. AI has
radically increased detection accuracy for complex data from the
clinic to self-driving cars, but often produces black-boxes that are
difficult to understand by users and have unknown failure
modes.
Augmented Intelligence will help designers make better and safer
decision systems by identifying problems with the data, such as
artefacts resulting in bias, but especially will help to integrate
analytical models into rational thinking, which involves more than
largely empirical performance measures. Hazard and Operability
studies, in particular, are crucial in safety-related applications
and require auditable tests of verification and validation for
software systems.
This special session will report methodologies and applications to
explain the operation of machine learning models. It will focus on
recent approaches to making the operation of machine learning
systems easier to understand such that users can usefully query the
operation of the model, changing any aspect to ensure safety in
operational environments.
Topics that are of interest to this session include but are not
limited to:
• Interpretable Machine Learning Models
• Visualisation of Data and Models
• Query Interfaces for Deep Learning
• Interactive User Interfaces
• Challenging applications of Active and Transfer
learning
• Relevance Learning and Metric Adaptation
• Practical Applications of Interpretable Machine
Learning
• Deep Neural Reasoning
Important Dates Paper
submission:December 15, 2018
Paper decision notification: January 30, 2019