https://popups.uliege.be/3041-539x/index.php?id=481 Index terms fr 0 https://popups.uliege.be/3041-539x/index.php?id=2153 The paper presents a methodology based on Artificial Neural Networks (ANN) to perform on-line a diagnosis of the health state of a machinery. The procedure at issue permits to detect the presence of backlash and to determine possible structural failures inside a mechanical system. Backlash and damages are important causes of vibrations in machines, therefore vibrations monitoring gives indirect information on these parasite effects. An ANN is used to classify the system behaviour among a predefined number of classes, receiving as input vibrational signals (simulated or measured). An application is discussed for devices purposely built for indexing motion, where compliance plays an important rôle, affecting the dynamic behavior of the whole machine. An analysis of parameters sensibility for the proposed procedure on simulated cases highlighted the best values and choices for these parameters. Tests of the procedure on experimental data collected on actual devices match closely the good results achieved with simulations. Tue, 30 Jul 2024 11:28:14 +0200 Thu, 10 Oct 2024 16:38:06 +0200 https://popups.uliege.be/3041-539x/index.php?id=2153 https://popups.uliege.be/3041-539x/index.php?id=1282 The integration between psychological concepts and neurophysiological data pose many epistemic difficulties which create almost unsolvable problems. We used the strategy of reconsidering experimental data and theoretical concepts within the frame of reference of some concepts from Quantum Physics and Chemistry. The model of a particle in a box against a potential barrier proved to be adequate to describe and interpret behavioural data from a neurophysiological viewpoint. Concerning cognitive processes we used the model of atom orbitals and of isolobal fragments to build a model of information processing in the frequency domain. After examining some problems related to anticipation, hypothesis construction and transformation we tested experimentally some of these ideas. Analysis of EEG records using Lee method of cross-correlation with Dirac delta waves repeated periodically yields distinctive spatial distributions of patterns of periodic waveforms for cognitive-affective states- joy, sadness, anxiety, anger and mistrust. It was possible, using exclusively electrophysiological indicators in the frequency domain, to classify adequately a group of 35 subjects that experienced these five cognitive-affective states. Wed, 10 Jul 2024 10:30:11 +0200 Thu, 10 Oct 2024 10:21:33 +0200 https://popups.uliege.be/3041-539x/index.php?id=1282 https://popups.uliege.be/3041-539x/index.php?id=3630 The paper describes specific current dynamic approaches to cognitive modelling alternative to the computational approach. Then a more general modelling framework is suggested that could accommodate both the computational and the dynamic approach, at least for what concerns the formalisms used. Thu, 26 Sep 2024 10:27:47 +0200 Thu, 10 Oct 2024 10:01:22 +0200 https://popups.uliege.be/3041-539x/index.php?id=3630 https://popups.uliege.be/3041-539x/index.php?id=643 Many analogies found in natural systems give evidence that the role of noise in a complex system might well lead to further organization. So, noise seems a good way in order to create novelty or to test the strength of algorithms. In this paper, we are going to analyse some artificial learning mechanisms such as genetic algorithms or neural networks, which may be generally formulated as an optimization problem by specifying a performance criterion, and then by using the simple but powerful technique of stochastic hill-climbing along the gradient. In these algorithms, the integration of random is a good way to maintain the exploration property during searching, useful for avoiding local optima or when environment is dynamic. We claim that artificial learning must overcome their limitations using the expedient of random search. This is due to attractors always present inside search procedures. We discuss in order to find another way to create order without having any presupposed attractors. This is also a central question for anticipatory systems which must learn about themselves and their environment. Fri, 28 Jun 2024 16:01:34 +0200 Tue, 08 Oct 2024 14:07:42 +0200 https://popups.uliege.be/3041-539x/index.php?id=643 https://popups.uliege.be/3041-539x/index.php?id=626 A two-level processing scheme for real-time image understanding is proposed, where an example-based (or case-based) reasoning in neural AI systems is introduced. The system has two levels; Component Level and Structure Level. At the component level, an elementary pattern recognition is performed as in the conventional pattern recognition, while the syntax pattern recognition is done at the structure level. Both levels are essentially time-consuming (theoretically, NP-complete each). The pattern recognition assisted by syntax recognition reduces the total complexity of processes, and the system can perform a real-time image understanding, when the VLSI chips are introduced. As a result, we show a reasonable real-time image understanding scheme by introducing neural pattern recognition at the component level and a case-based AI technique at the structure level. Fri, 28 Jun 2024 15:24:01 +0200 Tue, 08 Oct 2024 14:06:02 +0200 https://popups.uliege.be/3041-539x/index.php?id=626 https://popups.uliege.be/3041-539x/index.php?id=2605 This paper describes an approach to generate a sequence requiring an unrealizable function by programs, such as a flash that is required especially in creative activity of a human. We have already proposed a recurrent neural network that demonstrates a generation of several creative sequences, but convergency and stability problems occur. On the other hand, it is known in biological experiments where the chaotic sequences can be observed from brain waves. The neural network constructed from chaotic neurons has nonlinear dynamics, but there remains the difficulty of training method. We propose an evolutional methodology to train a chaotic neural network, and introduce Darwinism for its evolving process. To determine their most suitable structure and the weights of connection, we use AIC for the fitness value. Thu, 29 Aug 2024 15:07:30 +0200 Tue, 08 Oct 2024 14:04:11 +0200 https://popups.uliege.be/3041-539x/index.php?id=2605 https://popups.uliege.be/3041-539x/index.php?id=3572 The robotic inverse kinematic problem can be rightly classified as a very felt theme in the field of robotics. Many studies have been carried out in order to find new methods for the solution of the problem as alternatives to the traditional ones. In particular, every method able to improve the calculation speed is more and more appreciated. In the present paper an innovative method for the numerical inversion of non linear equations sets is shown. The approach is based on some procedures typical of the soft-computing area. In particular, the inverse kinematic problem is solved by a Neural Network optimised by means of a Genetic Algorithm acting inside an Hyperincursive scheme. After the introduction of the methodology developed, the paper shows some results obtained on a SCARA robot; they appear very good in terms of computational speed, even if the solution precision is not high near the boundaries of the working area. Thu, 26 Sep 2024 10:01:39 +0200 Tue, 08 Oct 2024 14:03:45 +0200 https://popups.uliege.be/3041-539x/index.php?id=3572 https://popups.uliege.be/3041-539x/index.php?id=477 In this paper a procedure to solve the problem of recognition and classification of sampled musical rythms is presented. The lack of precise rules for doing this analysis makes difficult and often ambiguous the automatic execution of a cognitive process naturally performed by human brain. This procedure can be extended to the classification of any signals showing similar characteristic (i.e. EEG or ECG). Due to the complexity of the time dependence, standard procedures used for chaos characterisation (i.e. correlation dimension, Lyapunov exponents, etc) can fail. Moreover a direct usage of artificial neural network can introduce too many optimization variables. The proposed procedure can be organized in two phases : the extraction of some new type of invariant from the sampled time series and the usage of this extracted features as input for a classifying standard neural network. This system was able to distinguish between binary and ternary signals with a precision of 99 %. The single rhythm was classified within an error of 5 %. This system seems to be able to deal with the behaviour that characterises a musical rhythmic sequence, and to classify patterns independently of the musical instrument and tempo. Thu, 27 Jun 2024 11:50:36 +0200 Tue, 08 Oct 2024 14:02:59 +0200 https://popups.uliege.be/3041-539x/index.php?id=477 https://popups.uliege.be/3041-539x/index.php?id=3601 We propose an operational research contribution to clinical psychopharmacology ; the key problems of drug selection and outcome prediction are tackled in a retrospective study about Mood and Anxiety Disorders focusing on sertraline, a selective inhibitor of serotonin reuptake. A three-step approach to data coding, clinical modeling and rule extraction is proposed, based on topographical techniques (Kohonen's Self-Organizing Maps) and information theory (Shannon entropy and mutual information). Clinical data are bitwise sampled, allowing an unbiased definition of system metrics. Uncertainty measures are introduced for a real-world sized approach to clinical practice; top-down induction decision trees (TDIDT) for drug administration are proposed, and a default logic of prescription is analyzed in the light of direct clinical experience and available literature data. Thu, 26 Sep 2024 10:13:22 +0200 Tue, 08 Oct 2024 13:28:18 +0200 https://popups.uliege.be/3041-539x/index.php?id=3601 https://popups.uliege.be/3041-539x/index.php?id=633 Experimental evidence suggests that modification of synaptic strength in the brain does not depend on co-activation of two connected neurons, as is assumed in most theoretical work since the proposals of Hebb (Hebb, 1949). Instead,through independent post-and presynaptic rules multiple modifications occur simultaneously at various sites in the nervous system. To account for this data, various researchers (Edelman, Fuster,...) propose an extension of the self-organising PDP approach to populational thinking. However, as in the PDP approach, the selection rules they propose only account for dynamical evolution of the system towards point attractors. The learning strategy of the networks is therefore still a purely bottom-up strategy. Experiments on visual perception seem to indicate that even low level visual processes can converge to more than one attractor (ambiguous figures, binocular rivalry), to limit cycles (oscillatory behaviour)or low-dimensional chaotic attractors. I argue to extend the neural network models of perceptual categorization to dynamical attractors and to include the multipticity of forms created by the autonomous, nonlinear brain dynamics as a complementary source of variation on which constraints of higher cognitive processes can act. Fri, 28 Jun 2024 15:36:56 +0200 Mon, 07 Oct 2024 15:29:14 +0200 https://popups.uliege.be/3041-539x/index.php?id=633