4 ms·
This was one of the most interesting quotes in the article: "The way to do this is by training the device: by running a task hundreds or perhaps thousands of t
by montecarl 9y ago
This was one of the most interesting quotes in the article:
"The way to do this is by training the device: by running a task hundreds or perhaps thousands of times, first with one type of input and then with another, and comparing which output best solves a task. “We don’t program the device but we select the best way to encode the information such that the [network behaves] in an interesting and useful manner,” Gimzewski said."
Since the "weights"/connections in their network are not easy to modify, they must instead figure out what kind of machine they have made by feeding it data and looking at the results. This does seem limiting in that the encoding might have to be arbitrarily complex to get the desired output.
- SilasX 9y agoSo ... a large part of the intelligence is actually coming from the user's decision of what encoding to use.
- ghgfhgfhn 9y agoYup, this is the principle behind reservoir computing. You take a complicated thing and hope that some of the processing that that complicated thing is doing, is useful to the task at hand. Then you set out to find that encoding/decoding. A nice way to think about it is - it's like taking the thing to be processed and smashing it into a million pieces and then searching through those broken pieces to assemble the thing you want to build. I am really pissed, because I've been wanting to do something like this for a decade. The interesting thing is that there are so many potential substrates for this kind of processing.