Contact:Elizna Huysamen
- 021 808 3244
Location: Van der Sterr building, 2nd Floor, Room 2048
Explainable Artificial Intelligence (XAI) is an emerging field of methods being developed to help interpretability for complex nonlinear models. At the core of XAI are local explanations, which provide insight into what factors most contribute to an individual prediction. Local interpretations examine the local neighbourhood of a predicted value. They can be converted into a linear combination of variables, which can then be interactively examined, against the predicted and observed values, using manual tour methods. This talk will describe how local explanations work, and also how to investigate the validity of any interpretations using visualisation.