ORIGINAL: fcharton
Hi LoBaron,
ORIGINAL: LoBaron
The complex problem is how to bring an AI to identify data needed to adress a specific task if the dataset
is not present from the beginning. So how do you enable an AI to both, create and categorize new datasets,
and successfully use the available data in a way it makes sense?
I believe an AI can do this, once it is given a metric that measures the similarity between datasets. With such a distance, you can build categories from scratch by clustering available data, and create new datasets by "averaging" similar ones.
In other words, the AI processes new datasets by combining old ones. This is very close to the way we learn by experience. We never see the same situation twice, but we can spot similarities between new problems and past ones, and react accordingly.
The difficulty, then, is to define a metric to classify events (and refine it using available data). At this point, there is a bad trade off, sometimes known as Bellman's curse.
- to correctly classify complex datasets, you need to take a lot of parameters into account,
- this means your distance (and your datasets) will "live" in a high dimension phase space
- but... the higher the dimension, the further apart two average data points,
- which makes it difficult to find similarities... and therefore learn...
Finding ways to build efficient algorithms in high dimensions is pretty much the subject of modern research on machine learning.
Hi Francois; That's pretty much where LTI is, as you may have guessed from the materials I sent you. We started very simple, with only an 18 element parametric space, but as soon as we expanded them upwards, convergence was problematic, to say the least. There seems to be an inherent instability in an open model. It's almost as though learning must be constrained to a rather truncated set of conditions, with proportional limitations. And that's yet another layer. Woof!!!
[ed] But this is such a totally bizarre place to be discussing things like this. It makes me wonder. If there's that much horsepower, why ain't they at the 'Tute AI site? Or Cal Tech's? Or Yale's, or Harvard's, or Chicago's, UCI's. Michigan's, or (damn me for saying it) USC's? There's a lot out there.
fcharton:
Do I understand this correct that we are (at least partly) facing the following issue here?
The human brain does what you wrote above. All the time. Without many difficulties.
This is what we probably call a creative process.
We want or need an AI to behave/develop in the same or similar way the human brain does.
The human brain developed an ability to classify datasets and combine them based on similarities, therefore learn and introduce new dataset
packages, over a long time evolutionary process governed by simple survival and chance to reproduce.
A brain with the ability to interconnect specific high dimension datasets if needed, even if they are separated -like the ability of the human brain-, had
advantages over a brain severely limited by the constraints when it comes to datapoints separated too far to find similarities and therefore learn.
The ability to connect far separated datapoints though introduces the danger of creating "white noise" or comparision of one datapoint with all other
datapoints on every single step, and so paralyses the process.
So the human brain achieved this ability over billions of years (starting with the first neurons clustered together) and the reason it exists and behaves in the way it
does is the result of trial and error, where failure meant extinction.
So to counter developing one of the failures we have to implement something similar from scratch into a "huge cluster of neurons". That means that we may have to
artificially create catalysers for such situations, probably common in the human brain, where they developed over multiple generations of natural selection, or
at least some steering mechanism that prevents such a "white noise" situation.
This somehow looks like part of the issue JWE describes.
