This paper describes the creation of a set of Matlab scripts that allow a researcher to efficiently program an FPGA to simulate a conductance-based neural network model. The researchers describe the use of their system to run a conductance-based model of 40 neurons with all-to-all connectivity at up to around 8x real-time (I wonder how long it takes to run the same model in software on a typical PC?).
Entrepreneur-turned-cognitive neuroscientist Jeff Hawkins is distributing a “research release” of their experimental code base implementing his idea of hierarchical temporal memory described in his book, “On Intelligence”. Hawkins drew inspiration for the model from his own reading about the structure and function of the human neocortex and believes that it represents the foundation for developing intelligent machines.
Jeff explains this surprising move to open source the code for the Numenta Platform for Intelligent Computing (NuPIC) on the Numenta web site:
Why are we making NuPIC available now?
We have been contacted by dozens of researchers and scientists who are excited about HTM and by our work at Numenta. These people are anxious to work on HTM, are willing to be pioneers, and are willing to accept the uncertainty associated with a new technology. We are making our tools available so that these sophisticated developers can start building a community around HTM technology. NuPIC has been under development for 18 months, is pretty solid, and is well documented – including several examples to make it easy to get started – so we’re ready to open up to more developers, even while knowing that we do not yet have benchmarking data, and we cannot make guarantees about applicability to specific problems.
Here’s why Hawkins thinks that HTMs are new.
We have been covering Hawkins’ work for a while now. See these previous posts for more background info.
Neurodudes is actively soliciting code reviews of the newly released software. Is NuPIC the next big thing, or are you left feeling cold? Post your thoughts yourself using the instructions on the right-hand column, or let us know at contactus -AT- neurodudes.com!
What is the right level of biological realism to model neuronal systems in order to understand their computational properties? Some recent papers may help shed some light on the subject. Models of the computational properties of local networks of neurons are starting to come into their own. This year has already seen at least two articles published in experimentalist journals based on the same core of theoretical work.
To bring you up to speed, I need to remind you what is going on in the world of experimental neuroscience.
Experimentalists are now able to record the single-cell activities of a whole population of neurons simultaneously. From Briggman, Abarbanel, Kristan (2006):
By using multi-electrode arrays or optical imaging, investigators can now record from many individual neurons in various parts of nervous systems simultaneously while an animal performs sensory, motor or cognitive tasks. Given the large multidimensional datasets that are now routinely generated, it is often not obvious how to find meaningful results within the data.
This paper goes on to provide a nice overview on mathematical methods that researchers are using to grapple with the challenge of understanding the dynamics of the neural systems they are recording from. They make the case that conceptual progress needs to be made on the interpretation of the data these results yield. How can we understand what computations these neurons are collectively performing?
(Incidentally, this topic is being explored in a conference happening this week at the Los Alamos National Laboratory, which, according to one of the conference session chairs, is intended to help shape future directions for the lab. Hopefully there will be webcasts from this conference.)
Postdoctoral/research scientist positions are available in the inter-disciplinary group of Dmitri Chklovskii at the new Janelia Farm Research Campus of the Howard Hughes Medical Institute located in the suburbs of Washington, D.C. Candidates are expected to have a PhD in neuroscience, physics, computer science or electrical engineering. Most of the work is theoretical or computational and is done in collaboration with several experimental laboratories. Successful applicants will work on projects centered on neuronal circuits such as high-throughput reconstruction of wiring diagrams as well as combining structural and physiological data to infer circuit function. Salary will be commensurate with qualifications. For more information about research directions in the group please see: http://www.hhmi.org/research/groupleaders/chklovskii.html
Interested applicants should send their CV and a statement of research interests to mitya (at) janelia.hhmi.org, and arrange for three recommendation letters to be emailed to me.
Computational neuroscience is a field where many successful researchers have a strong physics background. So far, the physics approach has provided a strong foundation from which to understand the brain. Recently, however, the influence of a computer science perspective has become more prominent. How can we understand the different perspectives that these disciplines bring to the field? Can we observe the influence of physics methodologies on the modern study of the brain? And if so, what is the consequence of our understanding of the brain through the lens of physics versus the lens of computer science?
One consequence may be the way that computational neuroscience models time in the brain. The study of physics generally conceptualizes time as continuous. Time is something to be plotted on the x-axis of a graph where some other quantity of interest is plotted on the y-axis.
In computer science, on the other hand, real time is rarely conceptualized explicitly. Computer scientists do not plot quantities against time unless they are profiling software for performance purposes, and even then, time is more generally thought of as number of operations. Thinking about operations generally leads computer scientists to think about time as discrete events.
I posit that the distinction between continuous and discrete time creates a foundational difference between the physics approach and the computer science approach to understanding how the brain works. Due to the discrete time conceptualization, computer scientists are more comfortable explaining the function of brain systems in terms of chains of events with definite beginnings and definite ends. Physicists, on the other hand, are more comfortable explaining the brain in terms of dynamics, which do not require definite beginnings or definite ends. Computer scientists care more what the consequence of an event is in the brain, whereas physicists are more concerned with an concise account of the dynamics of what is occurring.
This divide is visible in the distinct modeling approaches of neurons that derive from these two disciplines. The canonical neuronal model contributed by the physics philosophy is the multi-compartmental conductance based (Hodgkin-Huxley like) model. This model is concerned with matching waveforms of current and voltage traces with those that are measured in real neurons. This model helps us to understand how changes of the properties of excitable membranes over time result in changes of neuronal behavior over time. The computational complexity of these models is thought to prevent more than a few hundred neurons modeled in this way from being analyzed concurrently.
Alternatively, the canonical neuronal model contributed by the computer science philosophy is the integrate-and-fire neuron. This model does away with modeling conductances explicitly as functions of time and simply performs a weighted sum of its inputs at each time step. Here a time step is a discrete event whose duration is a parameter of the model. The simplicity of this model allows large networks to be constructed, which are useful for modeling systems of many thousands of neurons.
The physics approach provides insight into the activity of single cells and small networks, whereas the computer science approach provides insight into the activity of large networks. Neither approach is optimal and neither approach provides all the tools that are necessary to truly understand the brain. As these two perspectives are better understood, the field of computational neuroscience can benefit from finding creative ways to merge these two conceptions of time into models that capture both small scale and large scale neuronal activity.
In conclusion, I have demonstrated that what begins as a division between discrete and continuous time amounts to a divide between a bottom-up and a top-down approach. Furthermore, I have shown that understanding the relative contributions of different sciences to computational neuroscience is important for understanding the paradigms that pervade the field.
HOW DOES THE BRAIN CONTROL BEHAVIOR?
HOW CAN TECHNOLOGY EMULATE BIOLOGICAL INTELLIGENCE?
The conference is aimed at researchers and students of computational neuroscience, cognitive science, neural networks, neuromorphic engineering, and artificial intelligence. It includes invited lectures and contributed lectures and posters by experts on the biology and technology of how the brain and other intelligent systems adapt to a changing world. The conference is particularly interested in exploring how the brain and biologically-inspired algorithms and systems in engineering and technology can learn. Single-track oral and poster sessions enable all presented work to be highly visible. Three-hour poster sessions with no conflicting events will be held on two of the conference days. Posters will be up all day, and can also be viewed during breaks in the talk schedule.
ELEVENTH INTERNATIONAL CONFERENCE
ON COGNITIVE AND NEURAL SYSTEMS
May 16 – 19, 2007
Boston University
677 Beacon Street
Boston, Massachusetts 02215 USA
http://www.cns.bu.edu/meetings/
Sponsored by the Boston University
Center for Adaptive Systems and
Department of Cognitive and Neural Systems (http://www.cns.bu.edu/)
with financial support from the National Science Foundation (http://cns.bu.edu/CELEST/)
Dear Members,
I am a prospective graduate student interested in taking up Neural Engineering under EE or Biomedical Engg for research. But I have a lot of concerns and need help from a person who knows about the field well.
1. I have studied VLSI, DSP, Image Processing, Wireless Communication, Control Systems and Embedded Systems as graduate and undergraduate courses and have some research interest in Neural Networks and Machine Learning(That’s how I got interested in Neural Engg and Prosthetics). Which of these subjects will be of help in Neural Engg/Prosthetics research. Which will be of most relevance. Please list them in the order of relevance(high->low).
2. What are the applications of the research ?
3. What is the research and JOB scope for this field? Are there any companies who recruit people with this specialisation? How is the job scene in academia? How many univs are doing research in this field in US? Please let me know about the career progression in academia, like how much time does it take to get full time academic position after PhD?
4. Especially, what are the applications of this research in Robotics?
5. What are the current problems and research themes in universities?
6. What imaging technologies are used in this research?
Though my queries may seem a bit ameteuristic, it is very important for me to get clarity on these doubts.
Hope my queries will be answered.
Thanking all of you in advance,
sudhi
The October 6th issue of Science is a special issue devoted to computational neuroscience. From the introduction to the special issue:
Computational neuroscience is now a mature field of research. In areas ranging from molecules to the highest brain functions, scientists use mathematical models and computer simulations to study and predict the behavior of the nervous system. Simulations are essential because the present experimental systems are too complex to allow collection of all the data. Modeling has become so powerful these days that there is no longer a one-way flow of scientific information. There is considerable intellectual exchange between modelers and experimentalists. The results produced in the simulation lab often lead to testable predictions and thus challenge other researchers to design new experiments or reanalyze their data as they try to confirm or falsify the hypotheses put forward. For this issue of Science, we invited leading computational neuroscientists, each of whom works at a different organizational level, to review the latest attempts of mathematical and computational modeling and to give us an outlook on what the future might hold in store.
Of particular interest is a review article by Randall O’Reilly on biologically based computational models. He focuses on models of the pre-frontal cortex.
This past May, the Almaden Research center, part of IBM research, invited some provocative speakers on the topic of “Cognitive Computing” to come and speak. Since IBM recently invested a lot of money into understanding the brain with the Blue Brain project, it seems like this meeting was a way to figure out the next step along this path.
Powerpoint presentations and videos of the event are available online.
From the synopsis:
The 2006 Almaden Institute will focus on the theme of “Cognitive Computing” and will examine scientific and technological issues around the quest to understand how the human brain works. We will examine approaches to understanding cognition that unify neurological, biological, psychological, mathematical, computational, and information-theoretic insights. We focus on the search for global, top-down theories of cognition that are consistent with known bottom-up, neurobiological facts and serve to explain a broad range of observed cognitive phenomena. The ultimate goal is to understand how and when can we mechanize cognition.
Confirmed speakers include Toby Berger (Cornell), Gerald Edelman (The Neurosciences Institute), Joaquin Fuster (UCLA), Jeff Hawkins (Palm/Numenta), Robert Hecht-Nielsen (UCSD), Christof Koch (CalTech), Henry Markram (EPFL/BlueBrain), V. S. Ramachandran (UCSD), John Searle (UC Berkeley) and Leslie Valiant (Harvard). Confirmed panelists include: James Albus (NIST), Theodore Berger (USC), Kwabena Boahen (Stanford), Ralph Linsker (IBM), and Jerry Swartz (The Swartz Foundation).
In an impressive integrative effort, a new article in this month’s issue of Neuron describes a robust object and face classification model that is consistent with both behavioral and fMRI experiments.
From a preview of the article:
“A central theme that has emerged in research on face perception therefore is whether or not faces are “special” such that the cognitive and neural mechanisms that underlie their processing are different from those underlying the processing of other visual objects. [...] In this issue of Neuron, Jiang et al. (2006) provide a compelling array of evidence supporting the idea that the processing of faces and objects do not rely on qualitatively different mechanisms. In a series of experiments, Jiang et al. present and integrate findings from neural modeling, behavior, and fMRI, showing that face classification, similarly to object classification, can be achieved by a simple-to-complex architecture, based on hierarchical shape detectors. Furthermore, variations of this model can account for both configural and feature-based processing without qualitative modification of the model’s structure.”
The Riesenhuber lab, from which this work comes, has been working on object recognition in an integrative way. The lab is particularly “at the intersection of neuroscience and AI”.