Neuroinformatics Research in Vision
Department of Information & Computer Sciences
Toyohashi University of Technology
Tempaku, Toyohashi 441-8580, JAPAN
Funded By Science and Technology Agency of Japan
URL: http://www.neuroinformatics.nl/
A. Motivation
One of the frontiers of the 21st century is the elucidation of the complicated and elaborate functions of the brain, such as sensory perception, recognition, memory, emotion, etc. The specialization and segmentation of advanced research topics make it very difficult to integrate data so as to understand the functions of the brain. The introduction of information science technology in the analysis, processing, transmission, storage, integration, and utilization of information is indispensable. In particular, mathematical models are used to describe and integrate data and results obtained from a number of research fields. These mathematical models can be regarded as the platform that supports the simulation experiment indispensable for studying and understanding the function and mechanism of the human brain.
In general, Neuroinformatics should construct the support environment that integrates the databases devoted to various research fields with the data analysis techniques. This should promote research relying on mathematical models, and advance the comprehension and knowledge of brain neuronal systems. If the research of each conventional field is a warp, neuroinformatics is the woof which links them. Neuroinformatics is the new research approach for the twenty-first century, which fuses the experimental techniques with the mathematical and information science techniques.
The NRV project (Japanese Neuroinformatics Research in Vision) is the first project in Japan started in 1999 under the Science and Technology Agency of Japan, aimed at building the foundation of neuroinformatics research. Because of the wealth of data on the visual system, the NRV project will use vision research to promote experimental, theoretical and technical research in neuroinformatics. The first goal of the project is to construct mathematical models for each level of vision (single neuron, retinal neural circuit, visual function). The second goal is to build resources for neuroinformatics utilizing information science technologies and the research support environment which integrate them. The third goal is to realize a new vision device based on the brain-type information processing principle.
The NRV Project links the following research groups:
| G1: Modeling a single neuron by mathematical reconstruction |
| G2: Realization of virtual retina based on cell physiology |
| G3: Study on the visual function by computational and systems approaches |
| G4: Realization of artificial vision devices |
| G5: Fundamental neuroinformatics research and development of public neuro-informatics workbench |
Figure 1: NRV platform image
B. Results
Since the project has just started, we are now constracting the web page which includes results of each members scientific materials. At this moment, we announce on SATELLITE, which will be released as a free-software, in shortly.
C. Availabiliy
A partial goal of the project is to introduce SATELLITE (System Analysis Total Environment for Laboratory-Language and InTeractive Execution). SATELLITE is a software platform developed specifically for neural system analysis and modeling workbench in creating mathematical models of a single neuron,retinal circuits and higher-order visual functions. The basic concept of system analysis forms a cycle: data analysis, modeling, computer simulation, evaluation and experimental testing as shown in Figure 2. SATELLITE can handle a series of such research cycles interactively.
Figure 2: Concept of SATELLITE
SATELLITE consists of the SATELLITE-shell which provides interactive and C-like language processing system, and several modules which together cover more than 200 commands and (signal processing, numerical simulation, etc. See also Figure 3). The most important facility of SATELLITE-shell is an interactive operating environment. User can execute command sequence from the text file (batch processing) in case of the complex and large scale simulations. One can also visualize data and print it.
Platforms:
SATELLITE has been implemented and is fully operational on the following platforms:
| The Corresponding OS | |
| SunOS 4.1.x Solaris 2.5 and higher Digital UNIX 4.0A and higher (Tru64 UNIX) HP-UX 10.01 and higher IRIX 6.2 and higer FreeBSD 2.1.0 and higher Linux 2.0.0 and higer | |
| X Window Ver.11 R5 & OSF/Motif Ver.1.1 and higher versions |
ISPP (Interactive Signal Processing Package) is a module for data analysis utilizing signal processing and statistical theories. It represents an important tool for extracting the essential characteristics from the experimental data. Built-in commands and functions can be applied not only to time series, but also to multi-dimensional data. These are extremely important for extracting the characteristics from experimental data in modeling studies.
NCS (Neural Circuit Simulator) is a neural modeling and simulation environment. In this system, special description language is utilized to describe the neuronal properties and the network structure. This language creates an environment in which the large scale physiological model can be described easily.
NPE (Nonlinear Parameter Estimation module) is a facility for estimation of linear and nonlinear model parameters from experimental data. NPE can be applied to the models described by general programming language like C, Fortran and so on.
BPS (Back-Propagation Simulator) is devoted to examining neural network characteristics and capabilities. It incorporates effective tools for monitoring and analysis of the learning process and its dynamics. Functions for tracing synaptic weight changes offers precious data for analysis of learning process, local minima and internal representation.
Figure 3: SATELLITE module structure
D. HTML
http://www.neuroinformatics.gr.jpE. Contact(e-mail) address.
