Effecting modulation by global scalar values in a spiking neural network
Methods and apparatus are provided for effecting modulation using global scalar values in a spiking neural network. One example method for operating an artificial nervous system generally includes determining one or more updated values for artificial neuromodulators to be used by a plurality of entities in a neuron model and providing the updated values to the plurality of entities.
1. A method for operating an artificial nervous system, comprising:
determining one or more updated values for artificial neuromodulators to be used by a plurality of artificial neurons;
storing the updated values locally as global values, at one of the plurality of artificial neurons, wherein:
the plurality of artificial neurons are connected, via one or more synapses, to the artificial neuron storing the updated values;
a type of each of the one or more synapses corresponds to a type of one of the artificial neuromodulators and is associated with a weight value; and
determining the one or more updated values comprises updating the corresponding stored value for the artificial neuromodulators by the weight value when a spike occurs on the corresponding synapse; and
providing the updated values to the plurality of artificial neurons.
2. The method of claim 1 , wherein neuron-type parameters specify how to apply the updated values for the neuromodulators on a per-input-channel-per-neuron-type basis.
3. The method of claim 1 , wherein the artificial neuromodulators correspond to at least one of norepinephrine, acetylcholine, dopamine, or serotonin.
4. The method of claim 2 , wherein each input channel of each type of the artificial neurons has a flag indicating whether the input channel should be scaled by the updated values.
5. The method of claim 1 , wherein operation of the artificial nervous system is based at least in part on spiking events.
6. An apparatus for operating an artificial nervous system, comprising:
a processing system configured to:
determine one or more updated values for artificial neuromodulators to be used by a plurality of artificial neurons;
store the updated values locally as global values, at one of the plurality of artificial neurons, wherein:
the plurality of artificial neurons are connected, via one or more synapses, to the artificial neuron storing the updated values,
a type of each of the one or more synapses corresponds to a type of one of the artificial neuromodulators and is associated with a weight value, and
determining the one or more updated values comprises updating the corresponding stored value for the artificial neuromodulators by the weight value when a spike occurs on the corresponding synapse; and
provide the updated values to the plurality of artificial neurons; and
a memory coupled to the processing system.
7. The apparatus of claim 6 , wherein neuron-type parameters specify how to apply the updated values for the neuromodulators on a per-input-channel-per-neuron-type basis.
8. The apparatus of claim 6 , wherein the artificial neuromodulators correspond to at least one of norepinephrine, acetylcholine, dopamine, or serotonin.
9. The apparatus of claim 7 , wherein each input channel of each type of the artificial neurons has a flag indicating whether the input channel should be scaled by the updated values.
10. The apparatus of claim 6 , wherein operation of the artificial nervous system is based at least in part on spiking events.
11. An apparatus for operating an artificial nervous system, comprising:
means for determining one or more updated values for artificial neuromodulators to be used by a plurality of artificial neurons;
means for storing the updated values locally as global values, at one of the plurality of artificial neurons, wherein:
the plurality of artificial neurons are connected, via one or more synapses, to the artificial neuron storing the updated values,
a type of each of the one or more synapses corresponds to a type of one of the artificial neuromodulators and is associated with a weight value, and
determining the one or more updated values comprises updating the corresponding stored value for the artificial neuromodulators by the weight value when a spike occurs on the corresponding synapse; and
means for providing the updated values to the plurality of artificial neurons.
12. The apparatus of claim 11 , wherein neuron-type parameters specify how to apply the updated values for the neuromodulators on a per-input-channel-per-neuron-type basis.
13. The apparatus of claim 11 , wherein the artificial neuromodulators correspond to at least one of norepinephrine, acetylcholine, dopamine, or serotonin.
14. The apparatus of claim 12 , wherein each input channel of each type of the artificial neurons has a flag indicating whether the input channel should be scaled by the updated values.
15. The apparatus of claim 11 , wherein operation of the artificial nervous system is based at least in part on spiking events.
16. A non-transitory computer-readable medium having instructions executable to:
determine one or more updated values for artificial neuromodulators to be used by a plurality of artificial neurons;
store the updated values locally as global values, at one of the plurality of artificial neurons, wherein:
the plurality of artificial neurons are connected, via one or more synapses, to the artificial neuron storing the updated values,
a type of each of the one or more synapses corresponds to a type of one of the artificial neuromodulators and is associated with a weight value, and
determining the one or more updated values comprises updating the corresponding stored value for the artificial neuromodulators by the weight value when a spike occurs on the corresponding synapse; and
provide the updated values to the plurality of artificial neurons.
17. The non-transitory computer-readable medium of claim 16 , wherein neuron-type parameters specify how to apply the updated values for the neuromodulators on a per-input-channel-per-neuron-type basis.
18. The non-transitory computer-readable medium of claim 16 , wherein the artificial neuromodulators correspond to at least one of norepinephrine, acetylcholine, dopamine, or serotonin.
19. The non-transitory computer-readable medium of claim 17 , wherein each input channel of each type of the artificial neurons has a flag indicating whether the input channel should be scaled by the updated values.
20. The non-transitory computer-readable medium of claim 16 , wherein operation of the artificial nervous system is based at least in part on spiking events.