Hopfield network

GPTKB entity

Statements (39)
Predicate Object
gptkbp:instanceOf recurrent neural network
gptkbp:activatedBy sign function
gptkbp:capacity approximately 0.15N for N neurons
gptkbp:energyFunction gptkb:Lyapunov_function
gptkbp:field gptkb:artificial_intelligence
gptkb:machine_learning
computational neuroscience
theoretical neuroscience
gptkbp:form gptkb:discrete-time_dynamical_system
continuous-time dynamical system
gptkbp:generalizes gptkb:Little-Hopfield_network
gptkb:continuous_Hopfield_network
gptkb:stochastic_Hopfield_network
gptkbp:hardware digital circuits
analog circuits
https://www.w3.org/2000/01/rdf-schema#label Hopfield network
gptkbp:influenced modern neural networks
optimization algorithms
associative memory models
gptkbp:introducedIn 1982
gptkbp:limitation cannot store correlated patterns well
limited storage capacity
spurious states
gptkbp:namedAfter gptkb:John_Hopfield
gptkbp:neuronsAre binary
gptkbp:relatedTo gptkb:Ising_model
Boltzmann machine
content-addressable memory
gptkbp:roadType fully connected
gptkbp:state stable states correspond to local minima of energy
gptkbp:trainingRule gptkb:Hebbian_learning
gptkbp:updateRule asynchronous
synchronous
gptkbp:usedFor optimization
associative memory
gptkbp:bfsParent gptkb:Energy-based_models_in_machine_learning
gptkb:convolutional_neural_network
gptkb:Boltzmann_machine
gptkbp:bfsLayer 5