Apparatus and method for partial evaluation of synaptic updates based on system events
Apparatus and methods for partial evaluation of synaptic updates in neural networks. In one embodiment, a pre-synaptic unit is connected to a several post synaptic units via communication channels. Information related to a plurality of post-synaptic pulses generated by the post-synaptic units is stored by the network in response to a system event. Synaptic channel updates are performed by the network using the time intervals between a pre-synaptic pulse, which is being generated prior to the system event, and at least a portion of the plurality of the post synaptic pulses. The system event enables removal of the information related to the portion of the post-synaptic pulses from the storage device. A shared memory block within the storage device is used to store data related to post-synaptic pulses generated by different post-synaptic nodes. This configuration enables memory use optimization of post-synaptic units with different firing rates.
1. A computerized spiking network apparatus, comprising:
a storage apparatus; and
a pre-synaptic unit in communication with a post-synaptic unit through a communication channel;
wherein said computerized spiking network apparatus is configured:
to store information related to a plurality of output pulses in said storage apparatus, said plurality of output pulses generated by said post-synaptic unit at a plurality of generation times between a triggering pulse and a system event; and
to evaluate a plurality of updates based at least in part on a plurality of intervals between said triggering pulse and each of said plurality of generation times;
wherein said system event enables removal of at least a portion of said information related to said plurality of output pulses from said storage apparatus.
2. The apparatus of claim 1 , wherein said triggering pulse is generated by said pre-synaptic unit, and is communicated through said communication channel.
3. A method of operating a communications channel coupled to a post-synaptic unit in a computerized spiking neuronal network, the method comprising:
modifying said communications channel based at least in part on a first group of pulses associated with said post-synaptic unit, said first group of pulses occurring between a preceding trigger communicated via said communication channel and a system event; and
maintaining said communication channel substantially unmodified between said system event and said preceding trigger.
4. A method of improving operation of a shared storage of a computerized network apparatus comprising at least one node coupled to a channel, the method comprising:
storing, in said shared storage, information related to a plurality of pulses associated with said at least one node; and
updating said channel in response to a system event by at least a plurality of updates based at least in part on a plurality of intervals between a trigger communicated through said channel, and said plurality of pulses;
wherein updating said channel enables removal of at least a portion of said information from said shared storage.