connectionism
E899009
approach in artificial intelligence
computational paradigm
model of cognition
theoretical approach in cognitive science
Connectionism is a cognitive science and artificial intelligence approach that models mental processes using networks of simple, interconnected units whose learning and behavior emerge from patterns of activation and weight adjustment.
Statements (51)
| Predicate | Object |
|---|---|
| instanceOf |
approach in artificial intelligence
ⓘ
computational paradigm ⓘ model of cognition ⓘ theoretical approach in cognitive science ⓘ |
| alsoKnownAs |
PDP
NERFINISHED
ⓘ
neural network approach ⓘ parallel distributed processing NERFINISHED ⓘ |
| associatedWork | Parallel Distributed Processing volumes NERFINISHED ⓘ |
| contrastsWith |
classical computationalism
ⓘ
rule-based models of cognition ⓘ symbolic AI ⓘ |
| coreIdea |
cognition arises from distributed representations
ⓘ
knowledge is stored in patterns of connection weights ⓘ mental processes are modeled as emergent from networks of simple units ⓘ processing occurs in parallel across many units ⓘ |
| emergedIn | 1980s ⓘ |
| fieldOfStudy |
artificial intelligence
ⓘ
cognitive science ⓘ computational neuroscience NERFINISHED ⓘ |
| influenced |
computational models of language
ⓘ
deep learning ⓘ models of memory ⓘ models of perception ⓘ modern neural network research ⓘ |
| influencedBy |
neuroscience
ⓘ
psychology ⓘ statistical learning theory ⓘ |
| learningMechanism |
Hebbian learning
ⓘ
backpropagation ⓘ error-driven learning ⓘ weight adjustment based on experience ⓘ |
| models |
language processing
ⓘ
learning and generalization ⓘ memory retrieval ⓘ pattern recognition ⓘ |
| notableProponent |
David E. Rumelhart
NERFINISHED
ⓘ
Geoffrey Hinton NERFINISHED ⓘ James L. McClelland NERFINISHED ⓘ |
| philosophicalIssue |
explanatory adequacy for higher-level cognition
ⓘ
relationship between distributed and symbolic representations ⓘ |
| typicalArchitecture |
autoassociative network
ⓘ
feedforward network ⓘ recurrent network ⓘ |
| typicalUnit | simple neuron-like processing unit ⓘ |
| usesConcept |
activation pattern
ⓘ
artificial neural network ⓘ connection weight ⓘ content-addressable memory ⓘ distributed representation ⓘ graceful degradation ⓘ learning rule ⓘ |
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.