Search

An approach to find the best patient-specific seizure detection system for nocturnal frontal lobe epilepsy (NFLE) is proposed. Data of several patients affected by NFLE, extracted from the European Epilepsy Database, have been used for this study. As every patient possesses different physiological characteristics, several simulations were performed in order to find the best features to be extracted from the best number of electroencephalogram (EEG) channels and to be inputted to two-dimensional self-organizing maps (SOM). The proposed method allows the definition of maps capable of presenting meaningful information on the actual brain state, revealing the mapping potential of clustering the data coming from seizure and non-seizure patterns; moreover, it also suggests to use SOM for early seizure detection. In fact, the temporal sequence of the patterns in an EEG recording can be projected on the SOM, obtaining a trajectory that describes the dynamics of the brain state as captured by the EEG. The analysis of this trajectory allows one to detect the seizure event, highlighting the possibility to promote therapies aimed at rapid and targeted disarming the seizures. A further contribution of the paper is to propose a methodology to be followed to customize the prediction model to a new patient. The performances of the proposed approach, in terms of sensitivity and specificity, have been compared with those of other well-established detection models proving it is a good alternative to early and accurately detect nocturnal frontal lobe epilepsy seizures.

Knowledge discovery consists of finding new knowledge from databases where dimension, complexity, or amount of data is prohibitively large for human observation alone. The need for efficient data visualization and clustering is often faced, for instance, in the analysis, monitoring, fault detection, or prediction of various engineering plants. In this paper, two clustering techniques, K-means and Self-Organizing Maps, are used for the identification of characteristic regions for plasma scenario in nuclear fusion experimental devices. The choice of the number of clusters, which heavily affects the performance of the mapping, is firstly faced. Then, the ASDEX Upgrade Tokamak high-dimensional operational space is mapped into lower-dimensional maps, allowing to detect the regions with high risk of disruption, and, finally, the current process state and its history in time are visualized as a trajectory on the Self-Organizing Map, in order to predict the safe or disruptive state of the plasma.

A fault diagnosis procedure for analog linear circuits is presented. It uses an off-line trained neural network as a classifier. The innovative aspect of the proposed approach is the way the information provided by testability and ambiguity group determination is exploited when choosing the neural network architecture. The effectiveness of the proposed approach is shown by comparing with similar work that has already appeared in the literature.

A general purpose implementation of the Tabu Search metaheuristic, called Universal Tabu Search, is used to optimally design a Locally Recurrent Neural Network architecture. Indeed, the design of a neural network is a tedious and time consuming trial and error operation that leads to structures whose optimality is not guaranteed. In this paper, the problem of choosing the number of hidden neurons and the number of taps and delays in the FIR and IIR network synapses is formalised as an optimisation problem whose cost function to be minimised is the network error calculated on a validation data set. The performances of the proposed approach have been tested on the design problem of a Neural Network controller of a Custom Power protection device.

A general purpose implementation of the tabu search metaheuristic, called Universal Tabu Search, is used to optimally design a locally recurrent neural network architecture. The design of a neural network is a tedious and time consuming trial and error operation that leads to structures whose optimality is not guaranteed. In this paper, the problem of choosing the number of hidden neurons and the number of taps and delays in the FIR and IIR network synapses is formalised as an optimisation problem, whose cost function to be minimised is the network error calculated on a validation data set. The performance of the proposed approach has been tested on the problem of modelling the dynamics of a non-isothermal, continuously stirred tank reactor, in two different operating conditions: when a first order exothermic reaction is occurring; and when two consecutive first order reactions lead to a chaotic behaviour. Comparisons with alternative neural approaches are reported, showing the usefulness of the proposed method.

This paper presents a neural network system for the diagnosis of analog circuits and shows how the performance of such a system can be affected by the choice of different techniques used by its submodules. In particular we discuss the influence of feature extraction techniques such as Fourier Transforms, Wavelets and Principal Component Analysis. The system uses several different power supplies and as many neural networks “in parallel”. Two different algorithms that can be used to combine the candidate sets produced by each network are also presented. The system is capable of diagnosing multiple faults even if trained on single ones.