stdp

[never-to-never]

Hello, I would like to ask which paper the implementation of stdp in nest-gpu is based on. Can you reply in your spare time? thanks

[never-to-never]

@jhnnsnk

[golosio]

Hi, you can have a look at Golosio, B., Tiddia, G., De Luca, C., Pastorelli, E., Simula, F., and Paolucci, P. S. (2021b). Fast simulations of highly-connected spiking cortical models using GPUs. Front. Comput. Neurosci. 15, 13. doi: 10.3389/fncom.2021.627620 (pg 3 1st column in the pdf), and also Morrison, A., Diesmann, M., and Gerstner, W. (2008). Phenomenological models of synaptic plasticity based on spike timing. Biol. Cybern. 98, 459–478. doi: 10.1007/s00422-008-0233-1 and Sboev, A., Vlasov, D., Serenko, A., Rybka, R., and Moloshnikov, I. (2016). On the applicability of STDP-based learning mechanisms to spiking neuron network models. AIP Adv. 6:111305. doi: 10.1063/1.4967353 Keep in mind that currently NEST GPU only supports the symmetric-nearest-neighbor spike pairing scheme (Morrison et al., 2008) and most importantly that the arrival times of the presynaptic spike and of the postsynaptic spike to the synapse site depends not only on the times when the spikes are generated but also on the axonal and dendritic delays. If the time difference between the presynaptic and postsynaptic spikes is smaller than any of these delays, then they may even affect the order of arrival of the spikes to the synaptic site. In the folder python/test_stdp/cases you can find a set of tests for all possible combinations, compared with theoretical predictions. The scripts are very simple and I think from them it is very clear how the implementation of STDP works. I also have a technical note about those tests, but since it is unpublished you should write to my email address and specify how you want to use them.

[never-to-never]

thank you for your help. I'm experimenting with nest-gpu running monkey brain model, but want to implement some STDP variants based on nest-gpu. I don't know which stdp variants you think are suitable for implementation on the current nest-gpu framework. Can you give some references? thanks @golosio @jhnnsnk

[never-to-never]

thank you for your reply. I'm experimenting with nest-gpu running monkey brain model, but want to implement some STDP variants based on nest-gpu. I don't know which stdp variants you think are suitable for implementation on the current nest-gpu framework. Can you give some references? thanks 

------------------ 原始邮件 ------------------ 发件人: "nest/nest-gpu" @.>; 发送时间: 2022年6月16日(星期四) 晚上7:37 @.>; @.**@.>; 主题: Re: [nest/nest-gpu] stdp (Issue #34)

Hi, you can have a look at Golosio, B., Tiddia, G., De Luca, C., Pastorelli, E., Simula, F., and Paolucci, P. S. (2021b). Fast simulations of highly-connected spiking cortical models using GPUs. Front. Comput. Neurosci. 15, 13. doi: 10.3389/fncom.2021.627620 (pg 3 1st column in the pdf), and also Morrison, A., Diesmann, M., and Gerstner, W. (2008). Phenomenological models of synaptic plasticity based on spike timing. Biol. Cybern. 98, 459–478. doi: 10.1007/s00422-008-0233-1 and Sboev, A., Vlasov, D., Serenko, A., Rybka, R., and Moloshnikov, I. (2016). On the applicability of STDP-based learning mechanisms to spiking neuron network models. AIP Adv. 6:111305. doi: 10.1063/1.4967353 Keep in mind that currently NEST GPU only supports the symmetric-nearest-neighbor spike pairing scheme (Morrison et al., 2008) and most importantly that the arrival times of the presynaptic spike and of the postsynaptic spike to the synapse site depends not only on the times when the spikes are generated but also on the axonal and dendritic delays. If the time difference between the presynaptic and postsynaptic spikes is smaller than any of these delays, then they may even affect the order of arrival of the spikes to the synaptic site. In the folder python/test_stdp/cases you can find a set of tests for all possible combinations, compared with theoretical predictions. The scripts are very simple and I think from them it is very clear how the implementation of STDP works. I also have a technical note about those tests, but since it is unpublished you should write to my email address and specify how you want to use them.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>