Knowledge Extraction From Evolving Spiking Neural Networks With Rank Order Population Coding

Soltic, Snjezana; Kasabov, Nikola

doi:10.1142/s012906571000253x

Cited by 62 publications

(31 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The neuron's connections are established using the RO rule for the output neuron to recognise this vector (frame, static pattern) or a similar one as a positive example. The weight vectors of the output neurons represent centres of clusters in the problem space and can be represented as fuzzy rules (Soltic and Kasabov, 2010).…”

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

“…Fig.2. Example of an eSNN for classification using population RO coding of inputs (Soltic and Kasabov, 2010). Each input is connected to several feature neurons representing overlapping Gaussian receptive fields and producing spikes according to how much the current input variable value belongs to the receptive field: the higher the membership degree -the earlier a spike is generated and forwarded to the output neurons for learning or recall.…”

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

“…While an individual output neuron represents a single input pattern, merged neurons represent clusters of patterns or prototypes in a transformed spatial -RO space. These clusters can be represented as fuzzy rules (Soltic and Kasabov, 2010) that can be used to discover new knowledge about the problem under consideration.…”

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

See 2 more Smart Citations

Dynamic evolving spiking neural networks for on-line spatio- and spectro-temporal pattern recognition

et al. 2013

Self Cite

View full text Add to dashboard Cite

On-line learning and recognition of spatio-and spectro-temporal data (SSTD) is a very challenging task and an important one for the future development of autonomous machine learning systems with broad applications. Models based on spiking neural networks (SNN) have already proved their potential in capturing spatial and temporal data. One class of them, the evolving SNN (eSNN), uses a one-pass rank-order learning mechanism and a strategy to evolve a new spiking neuron and new connections to learn new patterns from incoming data. So far these networks have been mainly used for fast image and speech frame-based recognition. Alternative spike-time learning methods, such as SpikeTiming Dependent Plasticity (STDP) and its variant Spike Driven Synaptic Plasticity (SDSP), can also be used to learn spatio-temporal representations, but they usually require many iterations in an unsupervised or semi-supervised mode of learning. This paper introduces a new class of eSNN, dynamic eSNN, that utilise both rank-order learning and dynamic synapses to learn SSTD in a fast, on-line mode. The paper also introduces a new model called deSNN, that utilises rank-order learning and SDSP spike-time learning in unsupervised, supervised, or semi-supervised modes. The SDSP learning is used to evolve dynamically the network changing connection weights that capture spatio-temporal spike data clusters both during training and during recall. The new deSNN model is first illustrated on simple examples and then applied on two case study applications: (1) moving object recognition using address-event representation (AER) with data collected using a silicon retina device; (2) EEG SSTD recognition for brain-computer interfaces. The deSNN models resulted in a superior performance in terms of accuracy and speed when compared with other SNN models that use either rank-order or STDP learning. The reason is that the deSNN makes use of both the information contained in the order of the first input spikes (which information is explicitly present in input data streams and would be crucial to consider in some tasks) and of the information contained in the timing of the following spikes that is learned by the dynamic synapses as a whole spatio-temporal pattern. The paper is published in Neural Networks, Elsevier, e- AbstractOn-line learning and recognition of spatio-and spectro-temporal data (SSTD) is a very challenging task and an important one for the future development of autonomous machine learning systems with broad applications. Models based on spiking neural networks (SNN) have already proved their potential in capturing spatial and temporal data. One class of them, the evolving SNN (eSNN), uses a one-pass rank-order learning mechanism and a strategy to evolve a new spiking neuron and new connections to learn new patterns from incoming data. So far these networks have been mainly used for fast image and speech frame-based recognition. Alternative spike-time learning methods, such as Spike-Timing Dependent Plasticity (STDP) and its variant S...

show abstract

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

Section: Evolving Spiking Neural Network (Esnn)mentioning

confidence: 99%

See 1 more Smart Citation

Dynamic evolving spiking neural networks for on-line spatio- and spectro-temporal pattern recognition

et al. 2013

Self Cite

View full text Add to dashboard Cite

show abstract

“…Extracting fuzzy rules from an eSNN would make the eSNN not only efficient learning models, but also knowledge-based models. A method was proposed in [26] and illustrated in Figure 22 …”

Section: Evolving Spiking Neural Network (Esnn) and Fuzzy Rule Extramentioning

confidence: 99%

The Evolution of the Evolving Neuro-Fuzzy Systems: From Expert Systems to Spiking-, Neurogenetic-, and Quantum Inspired

Kasabov

2013

On Fuzziness

Self Cite

View full text Add to dashboard Cite

This chapter follows the development of a class of intelligent information systems called evolving neuro-fuzzy systems (ENFS). ENFS combine the adaptive/evolving learning ability of neural networks and the approximate reasoning and linguistically meaningful explanation features of fuzzy rules. The review includes fuzzy expert systems, fuzzy neuronal networks, evolving connectionist systems, spiking neural networks, neurogenetic systems, and quantum inspired systems, all discussed from the point of few of fuzzy rule interpretation as new knowledge acquired during their adaptive/evolving learning. This review is based on the author's personal (evolving) research, integrating principles from neural networks, fuzzy systems and nature. Early work on the integration of neural networks and fuzzy systems for knowledge engineering: Neuro-fuzzy expert systemsThe seminal work by Lotfi Zadeh on fuzzy sets, fuzzy rules and intelligent systems [36][37][38] opened the field for the creation of new types of expert systems that combined the learning ability of neural networks, at a lower level of information processing, and the reasoning and explanation ability of fuzzy rule-based systems, at the higher level. An exemplar system is shown in Figure 22.1, where at a lower level a neural network (NN) module predicts the level of a stock index and a fuzzy reasoning module combines the predicted values with some macro-economic variables, using the following types of fuzzy rules [18]:IF AND THEN (22.1) These fuzzy expert systems continued the development of the hybrid NN-rulebased expert systems that used crisp propositional and fuzzy rules [13,15,17]. They represented a major topic at some conferences (Figure 22.2).

show abstract

“…-Cognitive and emotional robotics [8,64]; -Neuro-rehabilitation robots [110]; -Modelling finite automata [17,78]; -Knowledge discovery from SSTD [101]; -Neuro-genetic robotics [74]; -Modelling the progression or the response to treatment of neurodegenerative diseases, such as Alzheimer's Disease [94,64] - fig.20. The analysis of the obtained GRN model in this case could enable the discovery of unknown interactions between genes/proteins related to a brain disease progression and how these interactions can be modified to achieve a desirable effect.…”

Section: Snn Software and Hardware Implementations To Support Stprmentioning

confidence: 99%

Knowledge Extraction From Evolving Spiking Neural Networks With Rank Order Population Coding

Cited by 62 publications

References 33 publications

Dynamic evolving spiking neural networks for on-line spatio- and spectro-temporal pattern recognition

Dynamic evolving spiking neural networks for on-line spatio- and spectro-temporal pattern recognition

The Evolution of the Evolving Neuro-Fuzzy Systems: From Expert Systems to Spiking-, Neurogenetic-, and Quantum Inspired

Evolving Spiking Neural Networks and Neurogenetic Systems for Spatio- and Spectro-Temporal Data Modelling and Pattern Recognition

Contact Info

Product

Resources

About