Schematically, this is my pointpic.twitter.com/gWWaR82Nxj
Show this thread
Humans are not insects; the few megabytes of brain-related information in your DNA define *how to learn* but contain relatively little *knowledge* -- and only animal-like basics. (reasons: can't pass on knowledge via DNA; not enough space; evolutionary scales not long enough)
-
-
This Tweet is unavailable.
-
A baby needs to train the sensors during the first months (for seeing: focus, see colors, turn the image upside down).
End of conversation
-
-
-
Interesting! Can you share a source of this information. Human DNA vs Animal DNA and amount of info it may contain with evolution
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
It is likely the earlier brains (i.e. reptilian and early mammalian) are hired wired with knowledge. It's just the neo-cortex that has to relearn basic skill during infant hood. In short, it's not exactly a clean slate, but rather, one that is guided.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I think I disagree with that. Although there isn't explicit knowledge transfer, learning seems much more favorable in some specific areas than others.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I've seen a research paper skill transfer via dna
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Those "few megabytes" have evolved so that when combined with experience, learning is very fast. Expressing the right prior in the right representation doesn't need very many bits. This is radically different from existing ML.
-
I agree, but I'd go further. Our brain works well even in purely symbolic domains (abstract math) which evolution definitely didn't directly optimize for. By optimizing for learning quickly in empirical reality, the brain has the ability to generalize to strange domains.
- Show replies
New conversation -
-
-
i think it gets really interesting though when we start thinking groups of cognizers instead of just individuals. that's the where the magic is, more so than in a solitary squishy brain.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I like to think of the brain as a highly complex model with very carefully designed weights. DNA holds the well designed weights and architecture, while evolution provided the optimal algorithm. Meaning we have millennia of learning in our infancy already.
-
And while the architecture and algorithm are by fare superior to anything we have developed at the moment; the set of optimal weights for the inferior architectures and algorithms are being inferred by the algorithms with less data and time than evolution did with us.
End of conversation
New conversation -
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.