Have there been any discoveries through ML outputs that made us realize something new about nature/people that we didn't see due to human perception?
Human mistakes are understandable because our own mental models match the mental models operated by others humans. Meanwhile, ML algorithms share little resemblance to these mental models, even though they were trained from human-labeled data. They fail in inscrutable ways.pic.twitter.com/QM0LXZPHym
-
-
-
Turns out image classification is a lot easier than we thought and can be done without a high-level abstract understanding of an image. This holds true in general: human mental models are a lot more general and powerful than what's required for any specific task.
- Show replies
New conversation -
-
-
Does this figure come from a paper?
-
Deep Learning with Python, first edition, chapter 9
End of conversation
New conversation -
-
-
Machine learning lives chained to the wall of the cavern and we are a biased fire projecting fuzzy shadows.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Hi Francois, Hope your new son, his mother and you are doing ok. What do you think about this:https://youtu.be/d8nzFqoEOvE
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
is there anything pytorch van do but keras cant?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Does this mean bias in ML is because of bias labelled data but ML algo's can never be biased?
- Show replies
New conversation -
-
-
I like how this connects the explainability and the fairness ideas to explain that humans are in fact less explainable and less fair than AI As part of that and consequence of our biases, we consider AI to be less fair and less explainable, when in fact it is just not-humanpic.twitter.com/fVDNaUP8tk
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.