<p>With so much focus on model explainability, especially impressive attempts like - <a href="https://github.com/ModelOriented/DALEX" class="autolinkedURL autolinkedURL-url" target="_blank">github.com/ModelOriented/DALEX</a> - what happens when we ask for an explanation of a wrong model? If the NN's underlying equation is developed/reached-at is wrong (co-efs or form of polynomial or anything); then blindly/algorithmically explaining it is Machine Lying?</p>
<p>#doubt #machine-learning #ann</p>


With so much focus on model explainability, especially impressive attempts like - https://github.com/ModelOriented/DALEX - what happens when we ask for an explanation of a wrong model? If the NN's underlying equation is developed/reached-at is wrong (co-efs or form of polynomial or anything); then blindly/algorithmically explaining it is Machine Lying?

#doubt #machine-learning #ann

Is explaining wrong models, lying?

I'm a Mechanical Engineer (B. Tech.); I love maths and want a career more involved with pure or applied maths. So I am grappling with statistical learning and CS these days.
