Closed mrahtz closed 8 years ago
Thank you very much. I've merged it.
On Tuesday, September 13, 2016, Matthew Rahtz notifications@github.com wrote:
This is my old port of Brian 2. I haven't tested the classification accuracy, but it seems to be producing reasonable spike output. (It's not exactly the same output as the original Brian 1 version, but I'm guessing that's because of a difference in PoissonGroup implementation between
Brian 1 and 2 which means that the input is slightly different.)
You can view, comment on, or merge this pull request online at:
https://github.com/zxzhijia/Brian2STDPMNIST/pull/4 https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_zxzhijia_Brian2STDPMNIST_pull_4&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=Qaw_FnZd5Mb7idW_SUSdVMN8qygCNv1SufPTWP-0tp8&e= Commit Summary
- Remove trailing whitespace
- s/b/b2/g
- Remove old args to NeuronGroup
- Fix unit definition
- Fix threshold/reset
- Fix subgroup calculation
- Fix connections
- Don't load Brian 1
- Fix rate monitors
- Fix input rate
- Rename a couple more b. to b2.
- Fix another PopulationRateMonitor
- Start fixing second set of Connections
- continue fixing second connect
- Fix connections
- Fix STDP
- Fix connection plotting
- Fix rates setting
- Fix initial run command
- Hopefully fix weight normalisation
- Fix rates units
- Fix network setup
- Fix connection saving
- Fix refrac string
- Fix problem with delays
- Remove global preferences
- Fix theta import
- Fix deprecated pre/post syntax
- Remove real-time spike plotting
- Fix warnings about integration method
- Fix warning about rates
- Hopefully slightly speed up simulation
- Fix subplot plot numbering
- REALLY fix theta loading
- Fix refractoriness
- Fix monitors
- Remove extra spike monitor
- Fix refractoriness properly
File Changes
- M Diehl&Cook_spiking_MNIST_Brian1.py https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_zxzhijia_Brian2STDPMNIST_pull_4_files-23diff-2D0&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=CqkZlJ3l5blg5pKn7Lp5-aWkroyiEUBvJPQp1nCckhA&e= (345)
Patch Links:
- https://github.com/zxzhijia/Brian2STDPMNIST/pull/4.patch https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_zxzhijia_Brian2STDPMNIST_pull_4.patch&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=Uj9lO4rF77T5Bs8p6fk2Zp-ureN6avl2KXtMAfwTSm8&e=
- https://github.com/zxzhijia/Brian2STDPMNIST/pull/4.diff https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_zxzhijia_Brian2STDPMNIST_pull_4.diff&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=Vi29b6aPN6oEkwxQ7qL1ODETPDjDXOg0K-yu1iT-pYc&e=
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_zxzhijia_Brian2STDPMNIST_pull_4&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=Qaw_FnZd5Mb7idW_SUSdVMN8qygCNv1SufPTWP-0tp8&e=, or mute the thread https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ALPsQNtWzM-5FvaXBohEM8UyixoqboujY1ks5qpnQ1gaJpZM4J7fts&d=CwMCaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=X3YsXL1AZsKHgsh28HFY8A&m=tqMAtgClMAOntsgeG5XXkJnwtPTWhJPNgr7-d0oH9EM&s=Linc8A9BABS-ur7xFiVAc-03pzwPv1xwSTlNEGwy9kg&e= .
This is my old port of Brian 2. I haven't tested the classification accuracy, but it seems to be producing reasonable spike output. (It's not exactly the same output as the original Brian 1 version, but I'm guessing that's because of a difference in
PoissonGroup
implementation between Brian 1 and 2 which means that the input is slightly different.)