US 7,599,902 B2Granted

Analogical reasoning system

AssigneeHRL Laboratories, LLC

InventorsJason Fox, Chris Furmanski, Collin Green

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present invention relates to a general-purpose analogical reasoning system. More specifically, the present invention relates to a high-performance, semantic-based hybrid architecture for analogical reasoning, capable of finding correspondences between a novel situation and a known situation using relational symmetries, object similarities, or a combination of the two. The system is a high-performance symbolic connectionist model which multiplexes activation across a non-temporal dimension and uses controlled activation flow based on an analogical network structure. The system uses incremental inference to stop inference early for object correspondence, uses initial mappings to constrain future mappings, uses inferred mappings to synchronize activation, and independent mapping based on roles, superficial similarity, or composites.

Claims33

Dependent claims

A computerized system for analogical reasoning, comprising a data processor having a neural network that is configured to perform symbolic substitution by finding a correspondence between a target analog and a source analog to infer a solution to the target analog based on the source analog;

wherein each of the source analog and the target analog includes a plurality of nodes in the form of:
object/predicate units (OP units), where OP units are nodes that represent object nodes and predicate nodes independently of how they are bound together;
proposition units (Prop units), where Prop units are nodes that correspond to complete propositional statements to form a relational structure,
sub-proposition units (SP units), where SP units are nodes that represent sub-proposition units that are conjunctive representations of role-filler bindings, binding predicate nodes to object nodes;
and
semantic units, where semantic units are nodes in a semantic domain that relate to a corresponding predicate node and object node, forming predicate semantic nodes and object semantic nodes respectively, where semantic units represent the meaningfulness of predicate and object nodes and the degree of similarity between multiple predicate nodes or multiple object nodes,
and where the source analog and the target analog share the same semantic domain with at least some shared semantic units.

4 more elements

A method for analogical reasoning, comprising an act of using a data processor having a neural network to perform an act of symbolic substitution by finding a correspondence between a target analog and a source analog to infer a solution to the target analog based on the source analog;

and
wherein in the act of act of using a data processor having a neural network to perform an act of symbolic substitution by finding a correspondence between a target analog and a source analog to infer a solution to the target analog based on the source analog, each of the source analog and the target analog includes a plurality of nodes in the form of:
object/predicate units (OP units), where OP units are nodes that represent object nodes and predicate nodes independently of how they are bound together;
proposition units (Prop units), where Prop units are nodes that correspond to complete propositional statements to form a relational structure,
sub-proposition units (SP units), where SP units are nodes that represent sub-proposition units that are conjunctive representations of role-filler bindings, binding predicate nodes to object nodes;
and
semantic units, where semantic units are nodes in a semantic domain that relate to a corresponding predicate node and object node, forming predicate semantic nodes and object semantic nodes respectively, where semantic units represent the meaningfulness of predicate and object nodes and the degree of similarity between multiple predicate nodes or multiple object nodes,
and where the source analog and the target analog share the same semantic domain with at least some shared semantic units.

5 more elements

A computer program product for analogical reasoning, the computer program product comprising computer-readable instruction means stored on a computer readable medium for causing a computer to perform symbolic substitution by finding a correspondence between a target analog and a source analog to infer a solution to the target analog based on the source analog;

and
further comprising instruction means for causing a computer to generate a source analog and a target analog, wherein each of the source analog and the target analog includes a plurality of nodes in the form of:
object/predicate units (OP units), where OP units are nodes that represent object nodes and predicate nodes independently of how they are bound together;
proposition units (Prop units), where Prop units are nodes that correspond to complete propositional statements to form a relational structure,
sub-proposition units (SP units), where SP units are nodes that represent sub-proposition units that are conjunctive representations of role-filler bindings, binding predicate nodes to object nodes;
and
semantic units, where semantic units are nodes in a semantic domain that relate to a corresponding predicate node and object node, forming predicate semantic nodes and object semantic nodes respectively, where semantic units represent the meaningfulness of predicate and object nodes and the degree of similarity between multiple predicate nodes or multiple object nodes,
and where the source analog and the target analog share the same semantic domain with at least some shared semantic units.

5 more elements

Dependencies are inferred from the claim text.