Visualizing Network Connectivity with 8 October 2009 ConnPlotter ConnPlotter Party! Network Hans Ekkehard Plesser & Eilen Nordlie Diagrams Connectivity Pattern Norwegian University of Life Sciences Tables Simula Research Laboratory ConnPlotter RIKEN Brain Sciences Institute Perspectives 8 October 2009
Outline Happy Birthday, Neural Network Simulators! 8 October 2009 ConnPlotter Network Diagrams Party! Connectivity Pattern Tables Network Diagrams Connectivity Pattern ConnPlotter Tables ConnPlotter Perspectives Perspectives
8 October 2009 ConnPlotter Happy Birthday, Neural Network Simulators! Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
Network Simulation: 55 years! 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
B. G. Farley & A. W. Clark, 1954 ◮ Simulation of self-organized systems by digital computer 8 October 2009 ◮ MIT Memory test ConnPlotter computer ◮ 4096 16-bit words ◮ 90.000 fetch/add per Party! sec Network Diagrams ◮ 64 leaky I&F neurons Connectivity ◮ δ -synapses w/ delay Pattern Tables ◮ exponentially decaying ConnPlotter threshold Perspectives ◮ Gaussian noise (LFG) ◮ 75% connectivity ◮ Hebbian learning
First Neuron Class: 40 years! ◮ Lars Walløe, J. K. S. Jansen, Kirsten Nygaard ◮ A Computer Simulated Model of a Secondary Order Sensory Neuron 8 October 2009 ◮ Kybernetik 6 :130–141 (1969) ConnPlotter ◮ Model of neurons in dorsal spino-cerebellar tract ◮ Direct comparison to experimental data ◮ Implemented in Simula on a Univac 1107 Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
8 October 2009 ConnPlotter Network Diagrams Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
What makes science science? Refutable hypotheses Hypotheses must be stated with sufficient detail and 8 October 2009 ConnPlotter precision so that one can devise meaningful tests or counterexamples. Reproducible experiments Party! Network Experiments must be described and performed so carefully, Diagrams that others can reproduce them. Genuine failure to Connectivity Pattern reproduce results invalidates original findings. Tables ConnPlotter Perspectives Accumulation of knowledge Accumulation of knowledge through exchange, evolution and (sometimes) revolution of ideas.
What makes science science? Refutable hypotheses Hypotheses must be stated with sufficient detail and 8 October 2009 ConnPlotter precision so that one can devise meaningful tests or counterexamples. Reproducible experiments Party! Network Experiments must be described and performed so carefully, Diagrams that others can reproduce them. Genuine failure to Connectivity Pattern reproduce results invalidates original findings. Tables ConnPlotter Perspectives Accumulation of knowledge Accumulation of knowledge through exchange, evolution and (sometimes) revolution of ideas.
What makes science science? Refutable hypotheses Hypotheses must be stated with sufficient detail and 8 October 2009 ConnPlotter precision so that one can devise meaningful tests or counterexamples. Reproducible experiments Party! Network Experiments must be described and performed so carefully, Diagrams that others can reproduce them. Genuine failure to Connectivity Pattern reproduce results invalidates original findings. Tables ConnPlotter Perspectives Accumulation of knowledge Accumulation of knowledge through exchange, evolution and (sometimes) revolution of ideas.
What makes science science? Refutable hypotheses Hypotheses must be stated with sufficient detail and 8 October 2009 ConnPlotter precision so that one can devise meaningful tests or counterexamples. Reproducible experiments Party! Network Experiments must be described and performed so carefully, Diagrams that others can reproduce them. Genuine failure to Connectivity Pattern reproduce results invalidates original findings. Tables ConnPlotter Perspectives Accumulation of knowledge Accumulation of knowledge through exchange, evolution and (sometimes) revolution of ideas.
What do we need? 8 October 2009 ConnPlotter ◮ Reliable, ◮ Precise, Party! ◮ Expressive, Network Diagrams ◮ Easy-to-Use Connectivity ◮ means to visualize our models of neuronal networks. Pattern Tables ConnPlotter Perspectives
What do we need? 8 October 2009 ConnPlotter ◮ Reliable, ◮ Precise, Party! ◮ Expressive, Network Diagrams ◮ Easy-to-Use Connectivity ◮ means to visualize our models of neuronal networks. Pattern Tables ConnPlotter Perspectives
What do we need? 8 October 2009 ConnPlotter ◮ Reliable, ◮ Precise, Party! ◮ Expressive, Network Diagrams ◮ Easy-to-Use Connectivity ◮ means to visualize our models of neuronal networks. Pattern Tables ConnPlotter Perspectives
What do we need? 8 October 2009 ConnPlotter ◮ Reliable, ◮ Precise, Party! ◮ Expressive, Network Diagrams ◮ Easy-to-Use Connectivity ◮ means to visualize our models of neuronal networks. Pattern Tables ConnPlotter Perspectives
What do we need? 8 October 2009 ConnPlotter ◮ Reliable, ◮ Precise, Party! ◮ Expressive, Network Diagrams ◮ Easy-to-Use Connectivity ◮ means to visualize our models of neuronal networks. Pattern Tables ConnPlotter Perspectives
What do we have? 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
What can we do? ◮ Develop standards for symbols (eg Kitano et al, Nature Biotechnol 2005) 8 October 2009 ConnPlotter ◮ Draw network at different levels (from Nordlie et al, 2009) A B C Vp(v)LS(e) Vp(v)LS(i) Vp(v)L4(e) Vp(v)L4(i) V (v) V (h) p p Party! Vp(v)LI(e) Network R p Vp(v)LI(i) Diagrams Vp(v)LI(i) Connectivity T Pattern p Vp(h)LI(i) Tables Ret. ConnPlotter ◮ Rp Tp Perspectives ◮ Problems: ◮ How to generate automagically? ◮ Confusing line crossings
Dot doesn’t help . . . 8 October 2009 ConnPlotter Vp_hL23in Vp_hL4in Vp_vL4in Vp_vL23in TpTpRelay Vp_hL4pyr Vp_hL56in Vp_vL56in Vp_vL4pyr Party! Network Diagrams TpTpInter RpRpNeuron Vp_hL56pyr Vp_vL56pyr Tp Rp Connectivity Pattern Retina Vp_hL23pyr Vp_vL23pyr Tables Ret Vp_h Vp_v ConnPlotter Perspectives
8 October 2009 ConnPlotter Connectivity Pattern Tables Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
NEST Topology: Simple Layers 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives
Real networks: Complex Layers 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Tables ConnPlotter Perspectives From Hill & Tononi, J Neurophysiol, 2005, 93, 1671–1698
NEST Topology: Composite Layer Elements 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Tables ◮ Each color represents a neuron model ConnPlotter Perspectives ◮ Connections are made by specifiying entire layer and model to connect to/from
Populations, Groups, Projections Population Homogeneous group of neurons with 2D-layout 8 October 2009 ConnPlotter Group Collection of populations, e.g., a layer Projection Rule for connecting two populations Mask Only target population neurons Party! inside mask are connected Network Kernel Probability of connection Diagrams Synapse model Connectivity Pattern Tables 1 ConnPlotter 1 0.8 0.8 Perspectives 0.6 0.6 0.4 0.4 0.2 0 0.2 −2 0 3 2 0 1 0 2 −1 −3 −2
Connectivity Pattern Table (CPT) ◮ Connectivity matrix showing kernels & masks ◮ Intensity = weight × probability 8 October 2009 ConnPlotter A B IG RG IG RG IG IG Party! Network Diagrams Connectivity Pattern E E Tables ConnPlotter RG RG Perspectives I I E I E I
Aggregate CPTs ◮ Condense by combining across populations, synapse 8 October 2009 models, or both ConnPlotter A B IG RG IG RG IG Party! Network IG Diagrams RG Connectivity Pattern C Tables IG RG ConnPlotter IG RG Perspectives RG
Different synapse types ◮ Different colors ◮ Co-occurring types placed side-by-side 8 October 2009 ConnPlotter IG RG IG Party! Network Diagrams Connectivity Pattern E Tables ConnPlotter RG Perspectives I E I
Aggregate with synapse types IG RG IG 8 October 2009 ConnPlotter E RG I E I Party! A B C Network IG RG IG RG IG RG Diagrams Connectivity IG IG IG Pattern Tables ConnPlotter E RG Perspectives RG RG I E I
The Hill-Tononi Model . . . 8 October 2009 ConnPlotter Party! Network Diagrams Connectivity Pattern Vp_hL23in Vp_hL4in Vp_vL4in Vp_vL23in Tables ConnPlotter TpTpRelay Vp_hL4pyr Vp_hL56in Vp_vL56in Vp_vL4pyr Perspectives TpTpInter RpRpNeuron Vp_hL56pyr Vp_vL56pyr Tp Rp Retina Vp_hL23pyr Vp_vL23pyr Ret Vp_h Vp_v
. . . and as CPT Tp Rp Vp_h Vp_v Ret Ret 8 October 2009 Inter ConnPlotter Tp Relay Rp L23in L23pyr L4in Vp_h Party! L4pyr Network L56in Diagrams L56pyr Connectivity L23in Pattern L23pyr Tables L4in Vp_v L4pyr ConnPlotter L56in Perspectives L56pyr Inter Relay L23in L4in L56in L23in L4in L56in L23pyr L4pyr L56pyr L23pyr L4pyr L56pyr
Recommend
More recommend
