FIFTH system for general-purpose connectionist computation
This work addresses the challenge of making connectionist computation more practical and declarative for researchers and practitioners in AI and programming languages, though it appears incremental by building on existing theoretical frameworks.
The paper tackles the problem of formalizing connectionist computation by developing a framework based on information propagation networks with unbounded recursion, which is Turing-complete and allows connectionist computations to interface directly with high-level programming language semantics. It also solves difficult NP-hard search problems in programming uniformly and approximately achieves information-theoretic performance limits.
To date, work on formalizing connectionist computation in a way that is at least Turing-complete has focused on recurrent architectures and developed equivalences to Turing machines or similar super-Turing models, which are of more theoretical than practical significance. We instead develop connectionist computation within the framework of information propagation networks extended with unbounded recursion, which is related to constraint logic programming and is more declarative than the semantics typically used in practical programming, but is still formally known to be Turing-complete. This approach yields contributions to the theory and practice of both connectionist computation and programming languages. Connectionist computations are carried out in a way that lets them communicate with, and be understood and interrogated directly in terms of the high-level semantics of a general-purpose programming language. Meanwhile, difficult (unbounded-dimension, NP-hard) search problems in programming that have previously been left to the programmer to solve in a heuristic, domain-specific way are solved uniformly a priori in a way that approximately achieves information-theoretic limits on performance.