Visualization and knowledge discovery from interpretable models

Jul, 2020

Increasing number of sectors which affect human lives, are using Machine Learning (ML) tools. Hence the need for understanding their working mechanism and evaluating their fairness in decision-making, are becoming paramount, ushering in the era of Explainable AI (XAI). So, in this contribution we introduced a few intrinsically interpretable models which are also capable of dealing with missing values, in addition to extracting knowledge from the dataset and about the problem, and visualisation of the classifier and decision boundaries: angle based variants of Learning Vector Quantization. The performance of the developed classifiers were comparable to those reported in literature for UCI’s heart disease dataset treated as a binary class problem. The newly developed classifiers also helped investigating the complexities of this dataset as a multiclass problem

Link

@inproceedings{GhoshIJCNN2020,
title = {Visualization and knowledge discovery from interpretable models},
author = {Sreejita Ghosh and Peter Ti\~no and Kerstin Bunte},
booktitle = {International Joint Conference on Neural Networks ({IJCNN})},
booktitle2 = {, IEEE World Congress on Computational Intelligence},
booktitle3 = {2020 International Joint Conference on Neural Networks, {IJCNN} 2020, Glasgow, United Kingdom, July 19-24, 2020},
month = {July},
address = {Glasgow, United Kingdom},
pages = {1--8},
year = {2020},
organization = {IEEE},
abstract = {Increasing number of sectors which affect human lives, are using Machine Learning (ML) tools.     Hence the need for understanding their working mechanism and evaluating their fairness in decision-making, are becoming paramount, ushering in the era of Explainable AI (XAI).     So, in this contribution we introduced a few intrinsically interpretable models which are also capable of dealing with missing values, in addition to extracting knowledge     from the dataset and about the problem, and visualisation of the classifier and decision boundaries: angle based variants of Learning Vector Quantization.     The performance of the developed classifiers were comparable to those reported in literature for UCI’s heart disease dataset treated as a binary class problem.     The newly developed classifiers also helped investigating the complexities of this dataset as a multiclass problem},
doi = {10.1109/IJCNN48605.2020.9206702},
url = {https://doi.org/10.1109/IJCNN48605.2020.9206702},
biburl = {https://dblp.org/rec/conf/ijcnn/GhoshTB20.bib},
bibsource = {dblp computer science bibliography, https://dblp.org},
}