Comments on: What would be your ideal computational neurobiology curriculum? http://neurodudes.com/2004/02/10/what-would-be-your-ideal-computational-neurobiology-curriculum/ at the intersection of neuroscience and AI. Sat, 05 Jul 2008 12:38:35 +0000 http://wordpress.org/?v=2.5.1 By: Bayle http://neurodudes.com/2004/02/10/what-would-be-your-ideal-computational-neurobiology-curriculum/#comment-1200 Bayle Wed, 25 Jan 2006 02:34:49 +0000 http://s93794016.onlinehome.us/wordpress/?p=11#comment-1200 ok i'm adding a course on "nonlinear dynamics for neurons and networks" based on Eugene Izhikevich's new textbook, <a href="http://www.nsi.edu/users/izhikevich/publications/dsn.pdf" rel="nofollow">The Geometry of Excitability and Bursting</a> ok i’m adding a course on “nonlinear dynamics for neurons and networks” based on Eugene Izhikevich’s new textbook, The Geometry of Excitability and Bursting

]]> By: Bayle http://neurodudes.com/2004/02/10/what-would-be-your-ideal-computational-neurobiology-curriculum/#comment-39 Bayle Wed, 31 Dec 1969 19:00:00 +0000 http://s93794016.onlinehome.us/wordpress/?p=11#comment-39 Here's some of mine: * (core; pick 1 or both) Neural Coding I: Theory of neural codes A review of postulated neural coding mechanisms and how to detect them experimentally. Place codes, frequency codes, spike timing codes, sparse codes. Feature coding. Types of distributed coding (which have been suggested so far). Types of correlation. For each coding type, explores mathematical theory, and also what experimental data would suggest/confirm/disprove the presence of that code in a biological system. * (core; pick 1 or both) Neural Coding II: Which codes are used where Reviews various neural subsystems (i.e. auditory periphery, auditory cortex, retina, optic nerve, V1, MT, somatosensation in the periphery, primary sensory cortex, primary motor cortex, premotor cortex, thalamus, etc). For each system, covers what is know about the electrophysiological responses of the cells in various conditions. Covers what has been deduced about the neural coding in these areas, or, when nothing is known, what is suspected. * (core) Math for neurobiologists I, II, III Linear algebra. Using Matlab. Differential equations. Probability. Statistics. Nonlinear systems analysis. Information theory. Algebra & logic. Brief segment on metalogic and the theory of computation. * (elective) A.I. models of neural systems I Neural network models. Spiking neural network models. * (elective) A.I. models of neural systems II. Dimensionality reduction. Machine learning. Probablistic methods. * (core) Computational models of specific neural systems Visits a variety of neural systems and reviews the top 4 or so computational models that have been proposed for each one. Thalamus. Cortex in general. V1. A1. MT. Retina. Cerebellum. Basal ganglia. Hippocampus. etc. * (elective) Data structures What kinds of data representations have been proposed, for neural systems and for computer science? Feature vectors, clusters, graphs, trees, inference rules, hierarchial categorization, semilattices, category-based memory vs. instance-based memory. Conceptions of memory (Markov assumption, RAM, semantic memory, episodic memory, procedural memory). * (core) Theoretical models of computation with neural elements I: basic circuit elements Covers a variety of models that have been proposed for how neural systems compute. How might neurons compute boolean OR? Boolean AND? Subtract a time difference between signals? Store memory? Do a Fourier or a wavelet transform? Etc. * (core) Theoretical models of computation with neural elements II: Covers stuff like Hopfield nets, Boltzman machines, etc. Also, nonlinear dynamics models of neural computation, such as hyterisis for memory storage, etc. * (elective) Theoretical neurobiology: Making deductions from constraints rather than simulations. What can we say about computation in neural systems which is falsifiable? "Simulation is an activity doomed to success." A review of how to look for evidence to constrain computational models. * (elective) Simulating individual neurons Equations, techniques, and software packages to model individual neurons and events at the subcellular level (Ca influx, ion channels, second messenger cascades, etc). Hodgekin-Huxley, etc. * (elective) Simulating small networks of neurons Equations, techniques, and software packages to model groups of neurons and events at the network level (i.e. where we don't care what is going on inside each neuron, as long as we have an equation to describe when it fires (or in the case of graded transmission, what its voltage is)). * (elective) Computational models of plasticity * (core) Basic Neuroanatomy * (core) Basic electrophysiology * (core) Introduction to systems neurobiology * (elective) Introduction to cognitive neuroscience * (elective) The latest & greatest in experimental techniques for computational neurobiology * (elective) Quantitative data analysis techniques for computational neurobiology * (elective) Neurally-inspired engineering Robotic and other applications of ideas derived from neural systems. Here’s some of mine:

* (core; pick 1 or both) Neural Coding I: Theory of neural codes
A review of postulated neural coding mechanisms and how to detect them experimentally.

Place codes, frequency codes, spike timing codes, sparse codes. Feature coding. Types of distributed coding (which have been suggested so far). Types of correlation. For each coding type, explores mathematical theory, and also what experimental data would suggest/confirm/disprove the presence of that code in a biological system.

* (core; pick 1 or both) Neural Coding II: Which codes are used where
Reviews various neural subsystems (i.e. auditory periphery, auditory cortex, retina, optic nerve, V1, MT, somatosensation in the periphery, primary sensory cortex, primary motor cortex, premotor cortex, thalamus, etc). For each system, covers what is know about the electrophysiological responses of the cells in various conditions. Covers what has been deduced about the neural coding in these areas, or, when nothing is known, what is suspected.

* (core) Math for neurobiologists I, II, III
Linear algebra. Using Matlab. Differential equations. Probability. Statistics. Nonlinear systems analysis. Information theory. Algebra & logic. Brief segment on metalogic and the theory of computation.

* (elective) A.I. models of neural systems I
Neural network models. Spiking neural network models.

* (elective) A.I. models of neural systems II.
Dimensionality reduction. Machine learning. Probablistic methods.

* (core) Computational models of specific neural systems
Visits a variety of neural systems and reviews the top 4 or so computational models that have been proposed for each one. Thalamus. Cortex in general. V1. A1. MT. Retina. Cerebellum. Basal ganglia. Hippocampus. etc.

* (elective) Data structures
What kinds of data representations have been proposed, for neural systems and for computer science? Feature vectors, clusters, graphs, trees, inference rules, hierarchial categorization, semilattices, category-based memory vs. instance-based memory. Conceptions of memory (Markov assumption, RAM, semantic memory, episodic memory, procedural memory).

* (core) Theoretical models of computation with neural elements I: basic circuit elements
Covers a variety of models that have been proposed for how neural systems compute. How might neurons compute boolean OR? Boolean AND? Subtract a time difference between signals? Store memory? Do a Fourier or a wavelet transform? Etc.

* (core) Theoretical models of computation with neural elements II:
Covers stuff like Hopfield nets, Boltzman machines, etc. Also, nonlinear dynamics models of neural computation, such as hyterisis for memory storage, etc.

* (elective) Theoretical neurobiology: Making deductions from constraints rather than simulations.
What can we say about computation in neural systems which is falsifiable? “Simulation is an activity doomed to success.” A review of how to look for evidence to constrain computational models.

* (elective) Simulating individual neurons
Equations, techniques, and software packages to model individual neurons and events at the subcellular level (Ca influx, ion channels, second messenger cascades, etc). Hodgekin-Huxley, etc.

* (elective) Simulating small networks of neurons
Equations, techniques, and software packages to model groups of neurons and events at the network level (i.e. where we don’t care what is going on inside each neuron, as long as we have an equation to describe when it fires (or in the case of graded transmission, what its voltage is)).

* (elective) Computational models of plasticity

* (core) Basic Neuroanatomy

* (core) Basic electrophysiology

* (core) Introduction to systems neurobiology

* (elective) Introduction to cognitive neuroscience

* (elective) The latest & greatest in experimental techniques for computational neurobiology

* (elective) Quantitative data analysis techniques for computational neurobiology

* (elective) Neurally-inspired engineering
Robotic and other applications of ideas derived from neural systems.

]]> By: Neville http://neurodudes.com/2004/02/10/what-would-be-your-ideal-computational-neurobiology-curriculum/#comment-40 Neville Wed, 31 Dec 1969 19:00:00 +0000 http://s93794016.onlinehome.us/wordpress/?p=11#comment-40 One thing that really sticks out when reading your curriculum is the heavy inter-mixing of theory classes and experiment/lab classes. For some reason -- even in departments that do both competently -- this is always like breaking some taboo... like someone could never do both. gasp! Of course, most of the cutting-edge PIs out there are the ones who are doing precisely that. Sure, you could focus on one in grad school and then develop the other skill-set as a postdoc but I like this solution a lot better. Note to upstart neuroscience programs: Do this now and you'll snatch away all the good students before the others even know what hit 'em. One thing that really sticks out when reading your curriculum is the heavy inter-mixing of theory classes and experiment/lab classes. For some reason — even in departments that do both competently — this is always like breaking some taboo… like someone could never do both. gasp!

Of course, most of the cutting-edge PIs out there are the ones who are doing precisely that. Sure, you could focus on one in grad school and then develop the other skill-set as a postdoc but I like this solution a lot better.

Note to upstart neuroscience programs: Do this now and you’ll snatch away all the good students before the others even know what hit ‘em.

]]> By: Bayle Shanks http://neurodudes.com/2004/02/10/what-would-be-your-ideal-computational-neurobiology-curriculum/#comment-41 Bayle Shanks Wed, 31 Dec 1969 19:00:00 +0000 http://s93794016.onlinehome.us/wordpress/?p=11#comment-41 Actually after reading your comment I think my post is a little light on experimental courses. Maybe I'd add elective courses on "slice electrophysiology", "culture electrophysiology", "genetic neurobiological methods", "plasticity protocols", "monkey psychophysics", "multielectode techniques & data analysis", "how to design, read & interpret electrophysiological experiments", and "developmental methods". As you can see, I think I may be partial to electrophysiology. Actually after reading your comment I think my post is a little light on experimental courses. Maybe I’d add elective courses on “slice electrophysiology”, “culture electrophysiology”, “genetic neurobiological methods”, “plasticity protocols”, “monkey psychophysics”, “multielectode techniques & data analysis”, “how to design, read & interpret electrophysiological experiments”, and “developmental methods”. As you can see, I think I may be partial to electrophysiology.

]]>