Abstract not found

The aim of the paper is to introduce the scope and the problems of a new research area called Computational Neurogenetics (CNG), along with some solutions and directions for further research. CNG is concerned with the study and the development of dynamic neuronal models integrated with gene models. This area brings together knowledge from various science disciplines, such as computer and information science, neuroscience and cognitive study, genetics and molecular biology. A computational neurogenetic model is created to model a brain function or a brain disease manifestation, or to be used as a general mathematical model for solving complex scientific and engineering problems. The CNG area goes beyond modelling a simple relationship between a single gene and a single neuronal function or a neuronal parameter. It is the interaction between hundreds and thousands of genes in a neuron and their relationships with the functioning of a neuronal ensemble and the brain as a whole (e.g., learning and memory, speech and vision, epilepsy, mental retardation, aging, neural stem cells, etc.). CNG models constitute a second generation of neural network models that are closer to biological neural networks in their complex symbiosis of neuronal learning dynamics and molecular processes. Concrete models are presented as examples—evolving connectionist systems (ECOS) with evolutionary parameter optimisation and the CNG model of a class of spiking neural network ensembles (CNG-SNN).

How do the multiple specialized cortical areas in the brain interact with each other to give rise to an integrated behavior is a largely unanswered question. In this paper, I propose that such an integration can be understood under the framework of analogy, and that the thalamus and the thalamic reticular nucleus (TRN) may be playing a key role in this respect. The proposed thalamocortical model of analogy heavily depends on a diverse set of temporal parameters including axonal delay and membrane time constant, each of which is critical for the proper functioning of the model. The model requires a specific set of conditions derived from the need of the model to process analogies. Computational results with a network of integrate-and-fire (IF) neurons suggest that these conditions are indeed necessary, and furthermore, data found in the experimental literature also support these conditions. The model suggests that there is a very good reason for each temporal parameter in the thalamocortical network having a particular value, and that to understand the integrated behavior of the brain, we need to study these parameters simultaneously, not separately.

Computing complex spiking artificial neural networks (SANNs) on conventional hardware platforms is far from reaching real-time requirements. Therefore we propose a neuro-processor, called NeuroPipe-Chip, as part of an accelerator board. In this paper, we introduce two new concepts on chip-level to speed up the computation of SANNs. These concepts are implemented in a prototype of the NeuroPipe-Chip. We present the hardware structure of the prototype and evaluate its performance in a system simulation based on a hardware description language (HDL). For the computation of a simple SANN for image segmentation, the NeuroPipe-Chip operating at 100 MHz shows an improvement of more than two orders of magnitude compared to an Alpha 500 MHz workstation and approaches real-time requirements for the computation of SANNs in the order of 10 6 neurons. Hence, such an accelerator would allow for applications of complex SANNs to solve real-world tasks like real-time image processing. The NeuroPipe-Chip has been fabricated in an Alcatel 0.35-- m digital CMOS technology.

Competitive learning is a general technique for training clustering and classification networks. We have developed an 11-transistor silicon circuit, that we term an automaximizing bump circuit, that uses silicon physics to naturally implement a similarity computation, local adaptation, simultaneous adaptation and computation and nonvolatile storage. This circuit is an ideal building block for constructing competitive-learning networks. We illustrate the adaptive nature of the automaximizing bump in two ways. First, we demonstrate a silicon competitive-learning circuit that clusters one-dimensional (1-D) data. We then illustrate a general architecture based on the automaximizing bump circuit; we show the effectiveness of this architecture, via software simulation, on a general clustering task. We corroborate our analysis with experimental data from circuits fabricated in a 0.35-µm CMOS process.

How do the multiple specialized cortical areas in the brain interact with each other to give rise to an integrated behavior is a largely unanswered question. In this paper, I propose that such an integration can be understood under the framework of analogy, and that part of the thalamus and the thalamic reticular nucleus (TRN) may be playing a key role in this respect. The proposed thalamocortical model of analogy heavily depends on a diverse set of temporal parameters including axonal delay and membrane time constant, each of which is critical for the proper functioning of the model. The model requires a specific set of conditions derived from the need of the model to process analogies. Computational results with a network of integrate-and-fire (IF) neurons suggest that these conditions are indeed necessary, and furthermore, data found in the experimental literature also support these conditions. The model suggests that there is a very good reason for each temporal parameter in the thalamocortical network having a particular value, and that to understand the integrated behavior of the brain, we need to study these parameters simultaneously, not separately.

All Rights Reservediii

Abstract — In a particular case of behavioural model reduction by ANNs, a validity domain shortening has been found. In mechanics, as in other domains, the notion of validity domain allows the engineer to choose a valid model for a particular analysis or simulation. In the study of mechanical behaviour for a cantilever beam (using linear and non-linear models), Multi-Layer Perceptron (MLP) Backpropagation (BP) networks have been applied as model reduction technique. This reduced model is constructed to be more efficient than the non-reduced model. Within a less extended domain, the ANN reduced model estimates correctly the non-linear response, with a lower computational cost. It has been found that the neural network model is not able to approximate the linear behaviour while it does approximate the non-linear behaviour very well. The details of the case are provided with an example of the cantilever beam behaviour modelling.