Martin McGinnity

Evolutionary Design Of Spiking Neural Networks

Unlike traditional artificial neural networks (ANNs), which use a high abstraction of real neurons, spiking neural networks (SNNs) offer a biologically plausible model of realistic neurons. They differ from classical artificial neural networks in that SNNs handle and communicate information by means of timing of individual pulses, an important feature of neuronal systems being ignored by models based on rate coding scheme. However, in order to make the most of these realistic neuronal models, good training algorithms are required. Most existing learning paradigms tune the synaptic weights in an unsupervised way using an adaptation of the famous Hebbian learning rule, which is based on the correlation between the pre- and post-synaptic neurons activity. Nonetheless, supervised learning is more appropriate when prior knowledge about the outcome of the network is available. In this paper, a new approach for supervised training is presented with a biologically plausible architecture. An adapted evolutionary strategy (ES) is used for adjusting the synaptic strengths and delays, which underlie the learning and memory processes in the nervous system. The algorithm is applied to complex non-linearly separable problems, and the results show that the network is able to perform learning successfully by means of temporal encoding of presented patterns.

Fault diagnosis of electronic system using artificial intelligence

With increasing system complexity, shorter product life cycles, lower production costs, and changing technologies, the need for intelligent tools for all stages of a products lifecycle is becoming increasingly important. The purpose of this article is to give a brief review how AI has been used in the field of electronic fault diagnosis. Topics discussed include: rule-based diagnostic systems; model-based diagnostic systems; case-based reasoning (CBR); fuzzy reasoning and artificial neural networks (ANN); hybrid approaches; IEEE diagnostic standards and automated diagnostic tool future developments.

GPU implementation of spiking neural networks for color image segmentation

Spiking neural networks (SNN) are powerful computational model inspired by the human neural system for engineers and neuroscientists to simulate intelligent computation of the brain. Inspired by the visual system, various spiking neural network models have been used to process visual images. However, it is time-consuming to simulate a large scale of spiking neurons in the networks using CPU programming. Spiking neural networks inherit intrinsically parallel mechanism from biological system. A massively parallel implementation technology is required to simulate them. To address this issue, modern Graphic Processing Units (GPUs), which have parallel array of streaming multiprocessors, allow many thousands of lightweight threads to be run, is proposed and proved as a pertinent solution. This paper presents an approach for implementation of an SNN model which performs color image segmentation on GPU. This approach is then compared with an equivalent implementation on an Intel Xeon CPU. The results show that the GPU approach was found to provide a 31 times faster than the CPU implementation.

Online versus offline learning for spiking neural networks: A review and new strategies

Spiking Neural Networks (SNNs) are considered to be the third generation of neural networks, and have proved more powerful than classical artificial neural networks from the previous generations. The main reason for studying SNNs lies in their close resemblance with biological neural networks. However their applicability in real world applications has been limited due to the lack of efficient training methods. For training large networks on large data sets, online learning is the more natural approach for learning non-stationary tasks. In this paper, existing offline and online learning algorithms for SNNs will be reviewed, the issue that online learning algorithms for SNNs were less developed will be highlighted, and future lines of research related to online training of SNNs will be presented.