The generalization power of ML systems is sure as hell incredibly overhyped. Until we have some sort of system that can reason across conceptual domains (so, dealing with conceptual metaphor as one part) the idea of generalization in AI as a field will always be just hype.
I'd say ML is both overhyped and underrated. People overestimate the intelligence & generalization power of ML systems (ML as a magic wand), but underestimate how much can be achieved with relatively crude systems, when applied systematically (ML as the steam power of our era)
-
-
-
However, in a way, it doesn't even matter if we can get to that point anytime soon. Even just the current ML toolkits/stack and processes are amazing and useful for so many different areas. Steam power of this era indeed.
End of conversation
New conversation -
-
-
Preach!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This is literally what he just said
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Agreed, if you can constrain the state space of the problem (fixed inputs and outputs, like movie ratings & friend suggestions) ML is powerful. Open domain systems are still too fragile in many cases like chatbot assistant NLP.
-
Yup had built a chatbot using DL and it gave near human responses. Do you know what noobs had to say "It doesn't talk like Human"
End of conversation
New conversation -
-
-
I agree, but at the same time Google's AutoML is both fascinating and scary. It's reinforcement learning writing ML code. Do you think it could generalize to more complex type of code?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I tend to agree. ML is more like a "steam engine" technology rather than "electricity" in some sense.
-
Electricity can be thought of as a refined notion of a steam engine, a means to facilitate power. Now, ML is more of a steam engine, but concepts such as transfer learning are pushing it towards electricity.
- Show replies
New conversation -
-
-
This is especially true for ML in health. We underestimate ML for image analysis but overestimate ML for drug discovery.
#machinelearningThanks. 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.