No, stronger. It means the Halting Problem can be solved. Time to break out the bubbly! Or not.
The more I look into it, the more this Super-Turing business sounds like the computability-theory version of FTL neutrinos. According to Martin Davis, Siegelmann's paper involves neural nets with arbitrary real-numbered (as opposed to integer or rational) weights being able to recognize languages on the alphabet {a,b} that a Turing machine can't.
So you have this thing that can get uncomputable results because it's given uncomputable inputs! In other words, no more interesting, or practical, than attempting to do computation with the n-body problem, and almost certainly nothing we can actually build.
The more I look into it, the more this Super-Turing business sounds like the computability-theory version of FTL neutrinos. According to Martin Davis, Siegelmann's paper involves neural nets with arbitrary real-numbered (as opposed to integer or rational) weights being able to recognize languages on the alphabet {a,b} that a Turing machine can't.
So you have this thing that can get uncomputable results because it's given uncomputable inputs! In other words, no more interesting, or practical, than attempting to do computation with the n-body problem, and almost certainly nothing we can actually build.