Spiking neural network (original) (raw)
Las redes neuronales de impulsos (en inglés: spiking neural networks) son un tipo de redes neuronales artificiales más realistas que las redes neuronales artificiales clásicas, es decir, procesan la información de una forma más similar a las redes neuronales biológicas.
Property | Value |
---|---|
dbo:abstract | Gepulste neuronale Netze (kurz: SNN, englisch: Spiking neural networks) sind eine Variante künstlicher neuronaler Netzwerke, die näher an biologischen neuronalen Netzen sind als beispielsweise das mehrlagige Perzeptron. Gepulste neuronale Netze werden auch als Netze der dritten Generation bezeichnet. Das erste wissenschaftliche Modell von gepulsten neuronalen Netzen wurde 1952 von Alan Lloyd Hodgkin und Andrew Huxley eingeführt. Dieses Modell beschreibt, wie Aktionspotentiale starten und durch das Netz propagiert werden. Die Pulse werden im Allgemeinen jedoch nicht direkt von Neuron zu Neuron weitergeleitet, sondern über chemische Substanzen (sogenannte Neurotransmitter) im synaptischen Spalt. Die Komplexität und Vielfalt biologischer Neuronen haben zu einigen Neuronenmodellen geführt: Das Integrate-and-Fire-Neuron (1907), das FitzHugh-Nagumo-Modell (1961–1962) und das Hindmarsh-Rose-Modell (1984). Aus Sicht der Informationstheorie ist ein Modell gesucht, das erklärt, wie Informationen durch Pulse codiert und decodiert werden. So ist beispielsweise nicht abschließend geklärt, ob die Informationen durch die Feuerrate oder durch eine zeitliche Codierung übertragen werden. (de) Las redes neuronales de impulsos (en inglés: spiking neural networks) son un tipo de redes neuronales artificiales más realistas que las redes neuronales artificiales clásicas, es decir, procesan la información de una forma más similar a las redes neuronales biológicas. (es) Les réseaux de neurones à impulsions (SNN: Spike Neural Networks, en anglais) sont un raffinement des réseaux de neurones artificiels (ANN: Artificial Neural Network, en anglais) où l’échange entre neurones repose sur l’intégration des impulsions et la redescente de l’activation, à l’instar des neurones naturels. L’encodage est donc temporel et binaire. Le caractère binaire pose une difficulté de continuité au sens mathématique (cela empêche notamment l’utilisation des techniques de rétropropagation des coefficients - telle que la descente de gradient - utilisées classiquement dans les méthodes d'apprentissage). L’encodage en temps pose de même des problèmes d’interprétation. Ces inconvénients sont aussi des avantages dans une perspective spatio-temporelle : l’intégration limite l’activation aux neurones voisins (espace) et tolère la perte d’information (temps). Ces éléments rendent les SNN théoriquement plus puissants que d’autres types d’ANN globalement dérivés de l'algorithme du perceptron. Les difficultés d’implémentation matérielle et de méthode d’apprentissage semblent toutefois des freins à l’émergence des SNN; l'industrie présente cependant des solutions commerciales depuis 2017 avec diverses annonces évoquant un intérêt pratique en termes de complexité adressable et d'efficacité énergétique. (fr) Spiking neural networks (SNNs) are artificial neural networks that more closely mimic natural neural networks. In addition to neuronal and synaptic state, SNNs incorporate the concept of time into their operating model. The idea is that neurons in the SNN do not transmit information at each propagation cycle (as it happens with typical multi-layer perceptron networks), but rather transmit information only when a membrane potential – an intrinsic quality of the neuron related to its membrane electrical charge – reaches a specific value, called the threshold. When the membrane potential reaches the threshold, the neuron fires, and generates a signal that travels to other neurons which, in turn, increase or decrease their potentials in response to this signal. A neuron model that fires at the moment of threshold crossing is also called a spiking neuron model. The most prominent spiking neuron model is the leaky integrate-and-fire model. In the integrate-and-fire model, the momentary activation level (modeled as a differential equation) is normally considered to be the neuron's state, with incoming spikes pushing this value higher or lower, until the state eventually either decays or - if the firing threshold is reached - the neuron fires. After firing the state variable is reset to a lower value. Various decoding methods exist for interpreting the outgoing spike train as a real-value number, relying on either the frequency of spikes (rate-code), the time-to-first-spike after stimulation, or the interval between spikes. (en) Una rete neurale spiking, in sigla SNN (dall'inglese spiking neural network), è una rete neurale artificiale a impulso che tenta di mimare più realmente le reti neurali naturali. Oltre allo stato sinaptico e neuronale una rete di questo tipo incorpora anche il concetto di tempo nel suo modello operativo. L'idea è che i neuroni artificiali non attivino in automatico ognuno un ciclo di propagazione come nelle reti multistrato con percettrone, ma piuttosto quando un potenziale di membrana - una intrinseca qualità del neurone correlata alla carica della sua membrana elettrica - raggiunge uno specifico valore. Quando un neurone si attiva genera un segnale che viaggia verso altri neuroni, che a turno incrementano o decrementano i loro potenziali in accordo a questo segnale. Per le SNN, l'attuale livello di attivazione (modellato come una equazione differenziale) è normalmente considerato uno stato del neurone, che con impulsi in arrivo spinge questo valore più in alto e poi si attiva o decade nel tempo. Esistono vari "metodi di codifica" per interpretare l'uscita del "treno di impulsi" come numero reale, facendo affidamento sulla frequenza dei picchi o sul tempo tra i picchi, per codificare le informazioni. (it) Импульсная нейронная сеть (ИмНС, англ. Pulsed neural networks, PNN) или Спайковая нейронная сеть (СНН, англ. Spiking neural network, SNN) — третье поколение искусственных нейронных сетей (ИНС), которое отличается от бинарных (первое поколение) и частотных/скоростных (второе поколение) ИНС тем, что в нем нейроны обмениваются короткими (у биологических нейронов — около 1—2 мс) импульсами одинаковой амплитуды (у биологических нейронов — около 100 мВ).Является самой реалистичной, с точки зрения физиологии, моделью ИНС. (ru) |
dbo:thumbnail | wiki-commons:Special:FilePath/Pulsed_neuron_model.jpg?width=300 |
dbo:wikiPageExternalLink | http://icwww.epfl.ch/~gerstner/SPNM/SPNM.html http://www.sjriddix.com https://www.neurallead.com |
dbo:wikiPageID | 10159567 (xsd:integer) |
dbo:wikiPageLength | 31283 (xsd:nonNegativeInteger) |
dbo:wikiPageRevisionID | 1124973708 (xsd:integer) |
dbo:wikiPageWikiLink | dbr:Caltech dbr:PyTorch dbr:Python_(programming_language) dbr:Electrical_synapse dbr:Electrophysiology dbr:TrueNorth dbr:Biological_neuron_model dbr:Brian_(software) dbc:Computational_statistics dbr:Application_software dbr:Human_Brain_Project dbr:Perceptron dbr:Unity_(game_engine) dbr:University_of_California,_Santa_Cruz dbr:University_of_Manchester dbr:University_of_Massachusetts_Amherst dbr:Deep_learning dbr:Network_topology dbr:Neurotransmitter dbc:Articles_containing_video_clips dbr:Massively_parallel_computing dbr:Membrane_potential dbr:Norse_(neuron_simulator) dbr:Neuroinformatics dbr:École_Normale_Supérieure dbr:Operating_Model dbr:GENESIS_(software) dbr:Gradient_descent dbr:NEST_(software) dbr:Convolutional_neural_network dbr:Computational_neuroscience dbr:SpiNNaker dbr:Synapse dbr:Backpropagation dbc:Computational_neuroscience dbr:C++ dbr:C_Sharp_(programming_language) dbr:Action_potential dbr:Hebbian_theory dbr:Lasso_(statistics) dbr:Leaky_integrator dbr:Neuroethology dbr:Neuromorphic_engineering dbr:ARM_architecture dbr:Duke_University dbr:Central_nervous_system dbr:Biological_neural_network dbr:Differential_equation dbr:Hindmarsh–Rose_model dbr:Recurrent_neural_network dbr:Heidelberg_University dbr:Attractor_network dbr:Artificial_neural_network dbc:Artificial_neural_networks dbr:KTH_Royal_Institute_of_Technology dbr:CoDi dbr:Cognitive_architecture dbr:Cognitive_computer dbr:Cognitive_map dbr:Echo_state_network dbr:Hodgkin–Huxley_model dbr:Models_of_neural_computation dbr:Spike-timing-dependent_plasticity dbr:Artificial_neuron dbr:Milliwatt dbr:Neuron_(software) dbr:Yale_University dbr:Integrate-and-fire dbr:Von_Neumann_architecture dbr:Neural_coding dbr:Neural_decoding dbr:Neurogrid dbr:Unsupervised_learning dbr:Spiking_neuron_model dbr:Temporal_coding dbr:FitzHugh–Nagumo_model dbr:Thalamocortical_radiations dbr:Motion_perception dbr:Pulse-coupled_networks dbr:Systems_neuroscience dbr:Neural_correlate dbr:Neuromorphic_computing dbr:Spike_train dbr:File:Artificial_synapses_based_on_FTJs.png dbr:File:Core_Top-Level_Microarchitecture.png dbr:File:Neuron-to-neuron_mesh_routing_model.png dbr:File:Predicting_STDP_learning_with_ferroelectric_synapses.png dbr:File:Pulsed_neuron_model.jpg dbr:File:Real-time-classification-and-sens...piking-deep-belief-network-Movie1.ogv dbr:File:Spiking_Neural_Network_Controlled...ect_Navigate_in_an_Random_Terrain.ogv dbr:File:Unsupervised_learning_with_ferroelectric_synapses.png dbr:Izhikevich dbr:SnnTorch_(software) |
dbp:wikiPageUsesTemplate | dbt:Citation_needed dbt:Colend dbt:ISBN dbt:Main dbt:Missing_information dbt:More_citations_needed dbt:More_citations_needed_section dbt:Multiple_issues dbt:Reflist dbt:See_also dbt:Cols |
dct:subject | dbc:Computational_statistics dbc:Articles_containing_video_clips dbc:Computational_neuroscience dbc:Artificial_neural_networks |
rdf:type | owl:Thing yago:WikicatArtificialNeuralNetworks yago:WikicatNeuralNetworks yago:Abstraction100002137 yago:Communication100033020 yago:ComputerArchitecture106725249 yago:Description106724763 yago:Message106598915 yago:NeuralNetwork106725467 yago:Specification106725067 yago:Statement106722453 |
rdfs:comment | Las redes neuronales de impulsos (en inglés: spiking neural networks) son un tipo de redes neuronales artificiales más realistas que las redes neuronales artificiales clásicas, es decir, procesan la información de una forma más similar a las redes neuronales biológicas. (es) Импульсная нейронная сеть (ИмНС, англ. Pulsed neural networks, PNN) или Спайковая нейронная сеть (СНН, англ. Spiking neural network, SNN) — третье поколение искусственных нейронных сетей (ИНС), которое отличается от бинарных (первое поколение) и частотных/скоростных (второе поколение) ИНС тем, что в нем нейроны обмениваются короткими (у биологических нейронов — около 1—2 мс) импульсами одинаковой амплитуды (у биологических нейронов — около 100 мВ).Является самой реалистичной, с точки зрения физиологии, моделью ИНС. (ru) Gepulste neuronale Netze (kurz: SNN, englisch: Spiking neural networks) sind eine Variante künstlicher neuronaler Netzwerke, die näher an biologischen neuronalen Netzen sind als beispielsweise das mehrlagige Perzeptron. Gepulste neuronale Netze werden auch als Netze der dritten Generation bezeichnet. Aus Sicht der Informationstheorie ist ein Modell gesucht, das erklärt, wie Informationen durch Pulse codiert und decodiert werden. So ist beispielsweise nicht abschließend geklärt, ob die Informationen durch die Feuerrate oder durch eine zeitliche Codierung übertragen werden. (de) Les réseaux de neurones à impulsions (SNN: Spike Neural Networks, en anglais) sont un raffinement des réseaux de neurones artificiels (ANN: Artificial Neural Network, en anglais) où l’échange entre neurones repose sur l’intégration des impulsions et la redescente de l’activation, à l’instar des neurones naturels. L’encodage est donc temporel et binaire. Ces inconvénients sont aussi des avantages dans une perspective spatio-temporelle : l’intégration limite l’activation aux neurones voisins (espace) et tolère la perte d’information (temps). (fr) Spiking neural networks (SNNs) are artificial neural networks that more closely mimic natural neural networks. In addition to neuronal and synaptic state, SNNs incorporate the concept of time into their operating model. The idea is that neurons in the SNN do not transmit information at each propagation cycle (as it happens with typical multi-layer perceptron networks), but rather transmit information only when a membrane potential – an intrinsic quality of the neuron related to its membrane electrical charge – reaches a specific value, called the threshold. When the membrane potential reaches the threshold, the neuron fires, and generates a signal that travels to other neurons which, in turn, increase or decrease their potentials in response to this signal. A neuron model that fires at the (en) Una rete neurale spiking, in sigla SNN (dall'inglese spiking neural network), è una rete neurale artificiale a impulso che tenta di mimare più realmente le reti neurali naturali. Oltre allo stato sinaptico e neuronale una rete di questo tipo incorpora anche il concetto di tempo nel suo modello operativo. L'idea è che i neuroni artificiali non attivino in automatico ognuno un ciclo di propagazione come nelle reti multistrato con percettrone, ma piuttosto quando un potenziale di membrana - una intrinseca qualità del neurone correlata alla carica della sua membrana elettrica - raggiunge uno specifico valore. Quando un neurone si attiva genera un segnale che viaggia verso altri neuroni, che a turno incrementano o decrementano i loro potenziali in accordo a questo segnale. (it) |
rdfs:label | Gepulste neuronale Netze (de) Red neuronal de impulsos (es) Réseau de neurones à impulsions (fr) Rete neurale spiking (it) Spiking neural network (en) Импульсная нейронная сеть (ru) |
rdfs:seeAlso | dbr:Biological_neuron_model |
owl:sameAs | freebase:Spiking neural network yago-res:Spiking neural network wikidata:Spiking neural network dbpedia-de:Spiking neural network dbpedia-es:Spiking neural network dbpedia-fa:Spiking neural network dbpedia-fr:Spiking neural network dbpedia-it:Spiking neural network dbpedia-ro:Spiking neural network dbpedia-ru:Spiking neural network https://global.dbpedia.org/id/53nVz |
prov:wasDerivedFrom | wikipedia-en:Spiking_neural_network?oldid=1124973708&ns=0 |
foaf:depiction | wiki-commons:Special:FilePath/Artificial_synapses_based_on_FTJs.png wiki-commons:Special:FilePath/Core_Top-Level_Microarchitecture.png wiki-commons:Special:FilePath/Neuron-to-neuron_mesh_routing_model.png wiki-commons:Special:FilePath/Predicting_STDP_learning_with_ferroelectric_synapses.png wiki-commons:Special:FilePath/Pulsed_neuron_model.jpg wiki-commons:Special:FilePath/Unsupervised_learning_with_ferroelectric_synapses.png |
foaf:isPrimaryTopicOf | wikipedia-en:Spiking_neural_network |
is dbo:wikiPageDisambiguates of | dbr:SNN |
is dbo:wikiPageRedirects of | dbr:Spiking_neural_networks dbr:Spiking_neuron |
is dbo:wikiPageWikiLink of | dbr:List_of_computer_simulation_software dbr:Bio-inspired_computing dbr:Biological_neuron_model dbr:EDLUT dbr:Tim_Vogels dbr:Analog_computer dbr:Massimiliano_Versace dbr:Glossary_of_artificial_intelligence dbr:NEST_(software) dbr:Optical_neural_network dbr:Simon_Stringer dbr:Computational_neurogenetic_modeling dbr:Tensor_Processing_Unit dbr:Weebit_Nano dbr:William_J._Schwartz dbr:Leabra dbr:Liquid_state_machine dbr:Neuromorphic_engineering dbr:Recurrent_neural_network dbr:Artificial_neural_network dbr:CoDi dbr:Models_of_neural_computation dbr:Autapse dbr:BCPNN dbr:Wulfram_Gerstner dbr:SNN dbr:Mind_uploading dbr:Neurorobotics dbr:Types_of_artificial_neural_networks dbr:Spiking_neural_networks dbr:Spiking_neuron |
is rdfs:seeAlso of | dbr:Nervous_system_network_models |
is foaf:primaryTopic of | wikipedia-en:Spiking_neural_network |