SLIDE 1
06/04/18 1
2 06/04/18
Information Coding in the Hippocampus
Knierim, 2012
Spatial Coding in a Full-Scale Computational Model of the - - PowerPoint PPT Presentation
Spatial Coding in a Full-Scale Computational Model of the - - PowerPoint PPT Presentation
Spatial Coding in a Full-Scale Computational Model of the Hippocampus Ivan Raikov, Aaron Milstein, Grace Ng, Ivan Soltesz Department of Neurosurgery, Stanford University Blue Waters Symposium 2018 06/04/18 1 Information Coding in the
SLIDE 2
SLIDE 3
3 06/04/18
T
- wards a Complete Description of the Circuitry
Underlying Sharp Wave-Mediated Memory Replay
Colgin, 2017.
SLIDE 4
4 06/04/18
Understanding Neural Systems
1. Computational level: What is the task? What logic is used to solve it? 2. Algorithmic level: How is relevant information represented and manipulated to perform the task? 3. Implementational level: How does hardware implement the algorithms?
Marr, D.; Poggio, T. (1976).
SLIDE 5
5 06/04/18
Understanding Neural Systems
Investjgatjng hippocampal spatjal and temporal coding at scale Modeling hippocampal neuroanatomy and connectjvity structure Cell and circuit-level parameter optjmizatjon
SLIDE 6
6 06/04/18
Modeling the Hippocampus at Full Scale
Unique synthetic granule cell morphology 3D distribution of 1,000,000 granule cells Hippocampus of rat in 3D brain volume Axonal distributions
SLIDE 7
7 06/04/18
Modeling the Hippocampus at Full Scale
~1.2M neurons ~18B synapses
SLIDE 8
8 06/04/18
Simulation of hippocampal spatial navigation
SLIDE 9
9 06/04/18
Spatial coding of full-scale hippocampal simulation on Blue Waters
GC
Hz 5 s physical time / 2048 BW nodes for 3 hrs
SLIDE 10
10 06/04/18
Nested parallel multi-objective
- ptimization
https://github.com/neurosutras/nested
SLIDE 11
11 06/04/18
Realistic synaptic mechanisms compatible with full-scale simulations
Saturating AMPA-R + Facilitating NMDA-R
- - Blue Waters can simulate full-scale networks with unique synapses,
but this is an ineffjcient use of resources
- - new synapse mechanism implementation combines biophysical realism
with computational effjciency
SLIDE 12
12 06/04/18
Synaptic weight distributions
Buzsáki & Mizuseki
- Nat. Rev. Neurosci., 2014
SLIDE 13
13 06/04/18
A scalable framework for parallel IO and analysis of full-scale neuronal network structure
SLIDE 14
14 06/04/18
A scalable framework for generation, simulation, and analysis of full-scale neuronal network structure
SLIDE 15
06/04/18 15