There are so many bottlenecks in underlying data labels & built data properties. This is pretty much apparent in most DL works yet we don’t want to recognize that. Data needs proper understanding. Current scope doesn’t help for one-size-fits both in Dx or research in healthcare.
#blackbox #DeepLearning is not the solution. The chatbots are mirroring human biases. So just using historic data will not be enough. There needs to be audits on what the models are learning. We need to develop testing mechanisms.https://twitter.com/DrMurphy11/status/1170774525987971076 …
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Audits produce more data, tons of it and it can take significant time to discover what a model has learned.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
LIME for NLP?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm not surprised at all. This is what medical students were being taught not so long time ago, if not even now. Usually algorithms produce what they have been fed. Biased data, biased outcomes.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.