1 / 7

Brian

Brian. A clock driven simulator for spiking neural networks in Python. What? Why?. Simulator package for the Python programming language Much easier to learn and use (compared to things like Neuron, NEST, C++) Flexible Easy to extend. Brian in action. from Brian import *

duena
Download Presentation

Brian

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Brian A clock driven simulator for spiking neural networks in Python

  2. What? Why? • Simulator package for the Python programming language • Much easier to learn and use (compared to things like Neuron, NEST, C++) • Flexible • Easy to extend

  3. Brian in action from Brian import * dv = lambda v,ge,gi: (ge+gi-(v+49*mV))/(20*ms) dge = lambda v,ge,gi: -ge/(5*ms) dgi = lambda v,ge,gi: -gi/(10*ms) P = NeuronGroup(4000,model=(dv,dge,dgi),threshold=-50*mV, reset=-60*mV,init=(-60*mV,0*mV,0*mV)) Pe = P.subgroup(3200) Pi = P.subgroup(800) Ce = Connection(Pe, P, 'ge') Ci = Connection(Pi, P, 'gi') Ce.connectRandom(Pe, P, 0.02, weight=1.62*mV) Ci.connectRandom(Pi, P, 0.02, weight=-9*mV) M = SpikeMonitor(P,True) Mv = StateMonitor(P,'v',record=0) run(1*second) plot(Mv.times,Mv[0]) rasterPlot(M) show() Ci P Pi Pe Ce

  4. Brian in action 2 Vr = -70*mV Vt = -55*mV taum = 10*ms taupsp = 0.325*ms weight = 4.86 * mV dV = lambda V,x,y: (-(V-Vr)+x)*(1./taum) dx = lambda V,x,y: (-x+y)*(1./taupsp) dy = lambda V,x,y: -y*(1./taupsp)+25.27*mV/ms+(39.24*mV/ms**0.5)*xi P = NeuronGroup(N=1000, model=(dV,dx,dy),init=(0*volt,0*volt,0*volt), threshold=Vt,reset=Vr,refractory=1*ms) Pinput = PulsePacket(t=50*ms,n=85,sigma=1*ms) Pgp = [ P.subgroup(100) for i in range(10)] C = Connection(P,P,'y') for i in range(9): C.connectFull(Pgp[i],Pgp[i+1],weight) Cinput = Connection(Pinput,P,'y') Cinput.connectFull(Pinput,Pgp[0],weight) Mgp = [SpikeMonitor(p,record=True) for p in Pgp] Minput = SpikeMonitor(Pinput,record=True) monitors = [Minput]+Mgp P.V = Vr + rand(len(P)) * (Vt-Vr) run(100*ms) rasterPlot(showgrouplines=True,*monitors) Pulse Packet Pgp[0] Pgp[1] Pgp[9] Mgp[0] Mgp[1] Mgp[9]

  5. Features • Neuron models (DE, stochastic DE, ion channels, alpha functions, etc.) • Analysis and plotting with NumPy/PyLab • Unit checking • Interactive simulations with Python shell • Automatic optimisation (code generation)

  6. How it works • Model • Threshold • Reset • NeuronGroup • Connection • Network

  7. Things still to do • STDP • Speed – more automatic code generation • Distributed computing, GPU • Multi-compartmental models • Static analysis • Standardisation (PyNN, NeuroML, FacetsML) • Realtime plotting • Database

More Related