Black Boxes Are Not Required
Data Skeptic
English - June 05, 2020 19:59 - 32 minutes - 37.2 MB - ★★★★★ - 477 ratingsScience Technology machinelearning datamining datascience science skepticism statistics Homepage Download Apple Podcasts Google Podcasts Overcast Castro Pocket Casts RSS feed
Deep neural networks are undeniably effective. They rely on such a high number of parameters, that they are appropriately described as “black boxes”.
While black boxes lack desirably properties like interpretability and explainability, in some cases, their accuracy makes them incredibly useful.
But does achiving “usefulness” require a black box? Can we be sure an equally valid but simpler solution does not exist?
Cynthia Rudin helps us answer that question. We discuss her recent paper with co-author Joanna Radin titled (spoiler warning)…
Deep neural networks are undeniably effective. They rely on such a high number of parameters, that they are appropriately described as “black boxes”.
While black boxes lack desirably properties like interpretability and explainability, in some cases, their accuracy makes them incredibly useful.
But does achiving “usefulness” require a black box? Can we be sure an equally valid but simpler solution does not exist?
Cynthia Rudin helps us answer that question. We discuss her recent paper with co-author Joanna Radin titled (spoiler warning)…