Alberto Luise
abstract
This project aims to analyze the compromise of performance vs. explainabil- ity typical between Symbolic and Sub-symbolic approaches, tackling a dataset used to estimated the median value of owner-occupied homes. After comparing the performance of two interpretable and two non-interpretable models, their respective results will be explained using only the features of the model for the former, and LIME for the latter. These will be evaluated by human subjects in a double-blind trial.
outcomes