High-level Inferencing in a Connectionist Network

TLDR

Connectionist models have struggled to represent and apply general knowledge rules that require variables, a variable‑binding problem that limits their ability to perform high‑level inferencing for planning, reasoning, and natural language understanding. This paper introduces ROBIN, a structured neural network designed to perform high‑level inferencing by enabling variable bindings and rule application. ROBIN achieves this by using signatures—activation patterns that uniquely identify concepts bound to roles—allowing multiple role‑bindings to propagate in parallel for rule application and dynamic inference path instantiation, with a constraint‑relaxation process selecting among newly instantiated inferences within a connectionist semantic network. ROBIN successfully resolves multiple contextual constraints to select the best interpretation among several ambiguous inference paths, a task that is difficult even for symbolic models.

Abstract

Abstract Connectionist models have had problems representing and applying general knowledge rules that specifically require variables. This variable binding problem has barred them from performing the high-level inferencing necessary for planning, reasoning, and natural language understanding. This paper describes ROBIN, a structured neural network model capable of high-level inferencing requiring variable bindings and rule application. Variable bindings are handled by signatures—activation patterns which uniquely identify the concept bound to a role. Signatures allow multiple role-bindings to be propagated across the network in parallel for rule application and dynamic inference path instantiation. Signatures are integrated within a connectionist semantic network structure whose constraint-relaxation process selects between those newly-instantiated inferences. This allows ROBIN to handle an area of high-level inferencing difficult even for symbolic models, that of resolving multiple constraints from context to select the best interpretation from among several alternative and possibly ambiguous inference paths.