Department of Physiology, Development and Neuroscience Optimization of neuron models using grid computing Mike Vella Department of Physiology, Development.

Slides:



Advertisements
Similar presentations
Neurophysics Adrian Negrean - part 2 -
Advertisements

Dendritic computation. Passive contributions to computation Active contributions to computation Dendrites as computational elements: Examples Dendritic.
What does EEG actually measure?
GUM*02 tutorial session UTSA, San Antonio, Texas Parameter searching in neural models Mike Vanier, Caltech.
Optical controlling reveals time- dependent roles for adult-born dentate granule cells Yan Gu, Maithe Arruda-Carvalho, Jia Wang, Stephen R Janoschka, Sheena.
1 The length constant of the dendritic tree markedly effects passive conduction.
1Neural Networks B 2009 Neural Networks B Lecture 1 Wolfgang Maass
Rules for Information Maximization in Spiking Neurons Using Intrinsic Plasticity Prashant Joshi & Jochen Triesch { joshi,triesch
Parallelized Evolution System Onur Soysal, Erkin Bahçeci Erol Şahin Dept. of Computer Engineering Middle East Technical University.
1 HW/SW Partitioning Embedded Systems Design. 2 Hardware/Software Codesign “Exploration of the system design space formed by combinations of hardware.
Effects of Excitatory and Inhibitory Potentials on Action Potentials Amelia Lindgren.
Modeling The quadratic integrate and fire follows from a reduced form (1) where F(V) is a voltage dependant function which aims to capture the voltage.
COGNITIVE SCIENCE 17 The Electric Brain Part 1 Jaime A. Pineda, Ph.D.
Cognitive Neuroscience How do we connect cognitive processes in the mind with physical processes in the brain?
Applying Multi-Criteria Optimisation to Develop Cognitive Models Peter Lane University of Hertfordshire Fernand Gobet Brunel University.
Subgoal: conduct an in-depth study of critical representation, operator and other choices used for evolutionary program repair at the source code level.
Understanding and Managing WebSphere V5
H-1 Network Management Network management is the process of controlling a complex data network to maximize its efficiency and productivity The overall.
Basic Models in Theoretical Neuroscience Oren Shriki 2010 Integrate and Fire and Conductance Based Neurons 1.
Genie: A Genetic Placement Algorithm James P. Cohoon William D. Paris © 1987 IEEE.
Attention Deficit Hyperactivity Disorder (ADHD) Student Classification Using Genetic Algorithm and Artificial Neural Network S. Yenaeng 1, S. Saelee 2.
WAM-BAMM '05 Parameter Searching in Neural Models Michael Vanier California Institute of Technology.
Development in hardware – Why? Option: array of custom processing nodes Step 1: analyze the application and extract the component tasks Step 2: design.
Genetic Programming on Program Traces as an Inference Engine for Probabilistic Languages Vita Batishcheva, Alexey Potapov
A Genetic Algorithms Approach to Feature Subset Selection Problem by Hasan Doğu TAŞKIRAN CS 550 – Machine Learning Workshop Department of Computer Engineering.
The Measurement of Cell Capacitance in Multi-compartment Cells
Generator Potentials, Synaptic Potentials and Action Potentials All Can Be Described by the Equivalent Circuit Model of the Membrane PNS, Fig 2-11.
Lotte Ramekers. Research questions Introduction Models Methods Experiments and results Conclusions Questions.
Cristian Urs and Ben Riveira. Introduction The article we chose focuses on improving the performance of Genetic Algorithms by: Use of predictive models.
Soft Computing Lecture 18 Foundations of genetic algorithms (GA). Using of GA.
Introduction to Neural Networks. Neural Networks in the Brain Human brain “computes” in an entirely different way from conventional digital computers.
March 3rd, 2006 Chen Peng, Lilly System Biology1 Cluster and SGE.
Distributed Database Systems Overview
Fuzzy Genetic Algorithm
E-science grid facility for Europe and Latin America E2GRIS1 Gustavo Miranda Teixeira Ricardo Silva Campos Laboratório de Fisiologia Computacional.
Blue Brain Project Carlos Osuna, Carlos Aguado, Fabien Delalondre.
ELeaRNT: Evolutionary Learning of Rich Neural Network Topologies Authors: Slobodan Miletic 3078/2010 Nikola Jovanovic 3077/2010
1 Circuitscape Design Review Presentation Team Circuitscape Mike Schulte Sean Collins Katie Rankin Carl Reniker.
Functions of the nervous system The nervous system is responsible for: - sensory perception - cognitive functions - motor functions - regulatory functions.
A Production Scheduling Problem Using Genetic Algorithm Presented by: Ken Johnson R. Knosala, T. Wal Silesian Technical University, Konarskiego Gliwice,
Announcements: 1.TA Office hours: Mon 10am-12 Wed 12-1pm Room S Prerequisites BGYA01H & BGYA02H OR BGYA01Y.
Coevolutionary Automated Software Correction Josh Wilkerson PhD Candidate in Computer Science Missouri S&T.
Introduction CSE 1310 – Introduction to Computers and Programming Vassilis Athitsos University of Texas at Arlington 1.
Solving the Euclidean Non-Uniform Steiner Tree Problem Using a Genetic Algorithm Ian Frommer, Dept. of Mathematics, US Coast Guard Academy Bruce Golden,
Nens220, Lecture 11 Introduction to Realistic Neuronal Networks John Huguenard.
INFSO-RI Enabling Grids for E-sciencE Using of GANGA interface for Athena applications A. Zalite / PNPI.
A field of study that encompasses computational techniques for performing tasks that require intelligence when performed by humans. Simulation of human.
Synaptic Plasticity Synaptic efficacy (strength) is changing with time. Many of these changes are activity-dependent, i.e. the magnitude and direction.
Multi-cellular paradigm The molecular level can support self- replication (and self- repair). But we also need cells that can be designed to fit the specific.
An Evolutionary Algorithm for Neural Network Learning using Direct Encoding Paul Batchis Department of Computer Science Rutgers University.
Hirophysics.com The Genetic Algorithm vs. Simulated Annealing Charles Barnes PHY 327.
Evolutionary Computation Evolving Neural Network Topologies.
Simulation Environments for Neuroscience Presented by Ali Nadalizadeh – IPM – Summer 2009 به نام خداوند بخشنده مهربان.
Who cares about implementation and precision?
Simulation Tools in Neuroscience
A451 Theory – 7 Programming 7A, B - Algorithms.
Hodgin & Huxley The problem: Explain action potentials
A Distributed Two-Level PMU-Only Linear State Estimator
Sodium Entry during Action Potentials of Mammalian Neurons: Incomplete Inactivation and Reduced Metabolic Efficiency in Fast-Spiking Neurons  Brett C.
Evolution in materio: Initial Experiments In Liquid Crystal
Brief Intro to Python for Statistics
An Overview of Evolutionary Cellular Automata Computation
Effects of Excitatory and Inhibitory Potentials on Action Potentials
Principles Governing the Operation of Synaptic Inhibition in Dendrites
Volume 60, Issue 5, Pages (December 2008)
Synaptic integration.
Volume 95, Issue 5, Pages e4 (August 2017)
Resting Membrane Potential
Coevolutionary Automated Software Correction
Principles Governing the Operation of Synaptic Inhibition in Dendrites
Presentation transcript:

Department of Physiology, Development and Neuroscience Optimization of neuron models using grid computing Mike Vella Department of Physiology, Development and Neuroscience

Cortical pyramidal neurons Abundant in the cortex of virtually every mammal Found in structures associated with advanced cognitive function Display excitability, plasticity Dendritic domains with distinct synaptic inputs

Department of Physiology, Development and Neuroscience Pyramidal neurons are very complex Highly-elaborate dendritic morphology Voltage-dependent ion channels Ligand-dependent ion channels (receptors) Various uniform and non- uniform spatial distributions of all ion channels Set of information to be included in a model is large

Department of Physiology, Development and Neuroscience Single neuron multi-compartment models

Department of Physiology, Development and Neuroscience Why optimize? Single neuron models provide a basis for understanding cell and local circuit function Maximal conductances, compartment capacitances, channel distributions – the large number of parameters makes it difficult to “hand-tune” Provides a basis for understanding if ion channel characterisation is sufficient

Department of Physiology, Development and Neuroscience Genetic Algorithms Genetic algorithms (GAs) optimize solutions by mimicking evolution, this includes: Mutation Crossover Reproduction ECSPY

Department of Physiology, Development and Neuroscience Experimental and simulated data “Current clamp” measurements are used Records membrane potential through injecting fixed current into a cell through recording electrode

Department of Physiology, Development and Neuroscience Camgrid architecture

Department of Physiology, Development and Neuroscience Work flow

Department of Physiology, Development and Neuroscience Preparing code to run on CamGrid Python installed on most nodes, but I prepared pre-compiled version, with the right python version and all needed libraries (numpy, scipy, sqlite etc..) NEURON – prepared a pre-compiled “portable NEURON” Shell scripts

Department of Physiology, Development and Neuroscience Data management: SQLite File-based, embeddable database system Makes handing large amounts of data clean Integrates well with Python, C/C++ etc.. Makes life much easier

Department of Physiology, Development and Neuroscience Communication with execute node: Paramiko Paramiko – SSH2 Protocol for Python Wrote a library for basic CamGrid tasks (contact me if you want this,

Department of Physiology, Development and Neuroscience My CamGrid experience pros: Speedup ~ (number of execute nodes) / 2 =>300 day optimization takes ~ 1 week => 1 million simulations possible Great support cons: Takes time to learn Pre-compilation of code can be tricky Problem diagnosis harder

Department of Physiology, Development and Neuroscience Results Initial results show optimizer finds correct parameter set when tested against known solution

Department of Physiology, Development and Neuroscience Results Becomes less accurate as simulation evolves Good fit can have similar features – not necessarily identical voltage trace

Department of Physiology, Development and Neuroscience Conclusions Neuron optimization complements experiments for building accurate models of single cells Computationally intensive task suited to CamGrid Preparation of software to run on CamGrid can be difficult, but may be very worthwhile.