In Greek mythology, an oracle is a mystic soothsayer with supernatural powers. Check Delphi.
In mathematics, an oracle is an abstract know-it-all in recursion theory. Remember that.
In the formal negotiation space, an oracle is an integrated relational intelligence engine:
IRIE©
So how does this work?
Consider the architecture of a typical processing scheme that maps inputs onto certain values
derived from processed outputs as conditioned by the parameters of some predictive model:
So how does this work?
Consider the architecture of a typical processing scheme that maps inputs onto certain values
derived from processed outputs as conditioned by the parameters of some predictive model:
Those parameters are either algorithmic (fixed), or adjustable (tunable) hence trainable. The
latter are the sort of parameters that conventional neural networks exploit.
latter are the sort of parameters that conventional neural networks exploit.
Now imagine a related scheme in which the processor (encoder) and co-processor (decoder)
are instead “guided” by an oracle � and its dual ∗� respectively:
are instead “guided” by an oracle � and its dual ∗� respectively:
In math jargon, the parameters of the processor � and co-processor ∗� are now “recursive in the jump”.
IRIE© is such an oracle (and its dual).
The details of this arrangement are best presented in the context of a formal negotiation, e.g. a game such as contract bridge. To win a game is to optimize a negotiation among the players of that game. However, any truly effective optimization of this sort must rely somehow on sufficient access to game states, including information on the private (sub)states of each player in the game. Consequently, that information is necessarily incomplete for any individual player, since the private states of the other players are “hidden” from that individual, hence inaccessible.
That’s a strategy problem for negotiation theory in general.
The solution is quite simple: incorporate an oracle that provides a faithful representation of
those otherwise inaccessible states. A possible demonstration that this oracle exists “in principle” can be glimpsed by equating it with an optimization over all future states of the game. This approach relies on transferring the inaccessibility of other private states to the future, where those once inaccessible states become revealed through play over time … in the future.
those otherwise inaccessible states. A possible demonstration that this oracle exists “in principle” can be glimpsed by equating it with an optimization over all future states of the game. This approach relies on transferring the inaccessibility of other private states to the future, where those once inaccessible states become revealed through play over time … in the future.
Different spin, same magic.
Perhaps. Perhaps not.
Note that playing into the future is exactly what deep learning model training purports to do; but for sufficiently complex negotiations, this training simply cannot provide a sufficiently dense subset of possible plays. Consequently, no conventional machine learning system can qualify as an oracle.
Alternatively, imagine swallowing this essential inaccessibility in hyperbolic 3-space. There’s a lot of room in hyperbolic space…. Anyway, that’s the sort of thing that an oracle might do.
It’s what IRIEã actually does.
tsutaoka on October 26, 2021