Scheduled System Maintenance:
On May 6th, system maintenance will take place from 8:00 AM - 12:00 PM ET (12:00 - 16:00 UTC). During this time, there may be intermittent impact on performance. We apologize for the inconvenience.
By Topic

Computational Intelligence for Measurement Systems and Applications, 2004. CIMSA. 2004 IEEE International Conference on

Date 14-16 July 2004

Filter Results

Displaying Results 1 - 25 of 56
  • Unmixing low ratio endmembers through Gaussian synapse ANNs in hyperspectral images

    Publication Year: 2004 , Page(s): 150 - 154
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (700 KB) |  | HTML iconHTML  

    In this paper we considered the application of Gaussian synapse based artificial neural networks to the detection and unmixing of endmembers in cases where some of them are mixed in a low ratio within hyperspectral images. These networks and the training algorithm developed are very efficient in the determination of the abundances of the different endmembers present in the image using a very small training set that can be obtained without any knowledge on the proportions of endmembers present. The validation and test of these networks is carried out through their application to a benchmark set of artificially generated hyperspectral images containing five endmembers with spatially diverse abundances. As a second test, we applied the strategy to a real image and checked their behavior in regions where there were transitions between zones that were labeled differently and compared them to a hypothetical evolution of the spectrum from the endmember corresponding to one of the regions to the endmember of the other. A very good correspondence was found. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A comparison of neural networks architectures for geometric modelling of 3D objects

    Publication Year: 2004 , Page(s): 155 - 160
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (767 KB) |  | HTML iconHTML  

    This paper presents a critical comparison between three neural architectures for 3D object representation in terms of purpose, computational cost, complexity, conformance and convenience, ease of manipulation and potential uses in the context of virtualized reality. The models can be easily transformed in size, position and shape. Potential uses of the presented architectures in the context of virtualized reality are for the modeling of set operations and object morphing, for the detection of objects collision, for object recognition, object motion estimation and segmentation. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • CIMSA/2004 Session 10: Computational Intelligence in Control

    Publication Year: 2004 , Page(s): 161
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (391 KB)  

    First Page of the Article
    View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Blood pressure estimation using neural networks

    Publication Year: 2004 , Page(s): 21 - 25
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (583 KB) |  | HTML iconHTML  

    Oscillometry is an indirect method to determine blood pressure. An inflatable and debatable cuff is placed on arm to observe oscillations at different pressure levels. Thus, an envelope obtained from the oscillations is related to the blood pressure. In our work, we extract few features from the oscillometric waveforms, and estimate blood pressure using feedforward neural networks. Feature strength is evaluated by computing the standard deviation of the errors. The results are compared with the traditional maximum amplitude pressure algorithm. A large noninvasively collected database is used for this purpose. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Nuclear steam generator level control by a neural network-tuning 2-DOF PID controller

    Publication Year: 2004 , Page(s): 168 - 173
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (674 KB)  

    This paper focuses on the level control of a steam generator in a nuclear power plant. It is very difficult to effectively control the level of the nuclear steam generator, because of the swelling and shrinking caused by many kinds of disturbances, such as feed water rate, feed water temperature, main steam flow rate, and coolant temperature. Up to the present time, the PI controller has been used for the level control, owing to the easy control algorithms and the advantage which have been proven on the nuclear power plant. However, since there are problems with stability control during low power and start-up, only a highly experienced operator can operate during those procedures. A great deal of time and an expensive simulator is needed for the training of an operator. In addition to studying this problem, this paper has studied the tuning of a 2-DOF PID (two-degrees of freedom PID) controller by a neural network in the level control of the steam generator of a nuclear power plant, through the simulation and the experimentation of the model steam generator. Results obtained by simulation reveal the importance of the controller tuning in nuclear steam generator level control, and results obtained on the experimental steam generator suggest methods which can be used to reduce swelling and shrinking. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Real-time data analysis of action potentials

    Publication Year: 2004 , Page(s): 26 - 29
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (615 KB) |  | HTML iconHTML  

    In this paper an automated approach for the measurement of the electrical activity of a biological neural network is proposed. This method can be applied in the drug development process to verify the lead compounds of the high throughput screening with cell-based assays and there with reducing animal experiments. This verification is also called high content screening. To be able to detect and to evaluate action potentials, which mainly represent the electrical cell activity, neurons are cultured on a silicon sensor chip with integrated electronics and a multielectrode array (MEA). Due to the high parallelism of the measurement efficient and flexible algorithms are needed to assess and to classify the acquired data in real time. A system, consisting of a field programmable gate array (FPGA) and a digital signal processor (DSP) provide the required implementation platform. Filtering based on the discrete wavelet transform removes superimposed noise and low frequency disturbances from the neural signal. This analysis offers also a method to compute an adaptive threshold, which is essential for the detection process. Subsequently the measured data is classified to provide the user with a feedback of the experiment. First promising evaluation results from simulations and proof of concept hardware implementations can be presented. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • ICIMSA/2004 Session 8: Computational Intelligence: Theory and Optimization

    Publication Year: 2004 , Page(s): 120
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (391 KB)  

    First Page of the Article
    View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Financial prediction using modified probabilistic learning network with embedded local linear models

    Publication Year: 2004 , Page(s): 81 - 84
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (605 KB) |  | HTML iconHTML  

    In this paper, a model is proposed which combines multiple local linear models with a novel modified probabilistic neural network (MPNN). The proposed model is shown to provide improved regularization with reduced computation utilizing semiparametric model approach and efficient vector quantization of data space. In this paper, the proposed model is shown to generalize better with reduced variance and model complexity in short-term financial prediction application. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Artificial neural networks for meteorological nowcast

    Publication Year: 2004 , Page(s): 36 - 39
    Cited by:  Papers (2)  |  Patents (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (646 KB) |  | HTML iconHTML  

    Weather forecast are a typical problem where a huge amount of data coming from different types of sensors must be elaborated by means of complex, time-consuming algorithms. This work presents a new approach where the data fusion is performed with soft computing techniques. A statistical-neural system is used to "nowcast" meteorological data measured by a weather station. The system is able to forecast the evolution of these parameters in next three hours, giving precious indications about the possibility of rain, ice, and fog in next future. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • CIMSA/2004 Session 9: Computational Intelligence in Image and Vision

    Publication Year: 2004 , Page(s): 137
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (391 KB)  

    First Page of the Article
    View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Crosstalk-driven placement based on genetic algorithms

    Publication Year: 2004 , Page(s): 70 - 75
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (657 KB) |  | HTML iconHTML  

    Deep-Sub-Micron (DSM) technologies of 0.18 micron and below enable the integration of logical circuits having more than 10 million gates. In such a DSM technology, it's important to consider improving crosstalk noise at initial phase of layout design. In this paper, we proposed a novel crosstalk-driven placement algorithm. The proposed algorithm based on genetic algorithm (GA) has a two-level hierarchical structure. For selection control, new objective functions are introduced for improving crosstalk noise, reducing power consumption, improving interconnection delay and dispersing wire congestion. Studies on floor planning and cell placement have been reported as being applications of GA to the LSI layout problem. However, no studies have ever seen the effect of applying GA in consideration of power, delay and congestion. Results show improvement of 6.7% for crosstalk noise on average. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On-line fuzzy neural modeling with structure and parameters updating

    Publication Year: 2004 , Page(s): 127 - 132
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (633 KB) |  | HTML iconHTML  

    In this paper we propose a novel online clustering approach which can be applied in a general class of fuzzy neural networks. Both structure identification and parameters learning are online. The new clustering method for the structure identification can separate input-output data into different groups (rulenumber) by online input/output data. For the parameter learning, our algorithm has two advantages over the others. First, the normal methods for parameter identification are based on a fixed structure and whole data, for example ANFIS, but after clustering we know each group corresponds to one rule, so we train each rule by its group data, it is more effective. Second, we give a time-varying learning rate for the common used backpropagation algorithm, we prove that the new algorithm is stable and faster than backpropagation algorithm. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A rough-GA hybrid algorithm for rule extraction from large data

    Publication Year: 2004 , Page(s): 85 - 90
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (795 KB) |  | HTML iconHTML  

    The process of knowledge discovery from vast real life data is encountered with varieties of problems like, presence of noise and outliers in the data set, selection of proper subset of attributes (features) from a large number of relevant and irrelevant attributes, fuzzification or discretization of real-valued data, and finally rule induction. In this proposal, the process of rule creation has two steps. The first step consists of attribute selection, which is based on rough set theory. The next phase is to explore optimal set of simple yet accurate rules. This is accomplished by genetic algorithm. Here, the contribution is how to set the fitness of chromosomes so that simplicity-accuracy tradeoff is accomplished. Finally, chromosomes are coalesced to further simplify and reduce the number of rules. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Influence of forcings and circulation patterns on mean temperatures at different scales: an analysis by neural network modeling

    Publication Year: 2004 , Page(s): 51 - 56
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (713 KB) |  | HTML iconHTML  

    We present an analysis of the influence of various forcings and circulation patterns on annual and seasonal temperatures observed in the past, both at global and regional scales. In this framework, multilayer perceptrons show their ability to fully catch nonlinear relationships among these variables and allow us to "weight" the magnitude of different causes on the temperature behavior. In particular, our results show the necessity of including anthropogenic inputs for explaining the temperature behavior at global scale. Furthermore, we can assess the relative influences of global forcings and regional circulation patterns in determining regional temperature trends. Therefore, this activity can be very useful in order to identify the fundamental elements for a successful downscaling of Atmosphere-Ocean General Circulation Models, even on future scenarios. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Monitoring of natural scenes for feature extraction and tracking an independent component analysis (ICA) approach

    Publication Year: 2004 , Page(s): 2 - 3
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (521 KB) |  | HTML iconHTML  

    An independent component analysis (ICA) approach to monitoring of natural scenes empirically generates robust image features for localization and tracking of potentially-occluded targets. The ICA-based empirical model utilizes statistical techniques that assist analysts in characterizing the underlying criteria that enables such feature extraction. Thus, this approach provides a basis for analyzing how the empirically generated feature localization and tracking models and related algorithms to perform their function. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Simultaneous search for multiple routes using genetic algorithm

    Publication Year: 2004 , Page(s): 77 - 80
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (604 KB) |  | HTML iconHTML  

    Search for optimal route from source to destination is a well-known optimization problem and lot of good solutions like Dijkstra algorithm, Bellman-Ford algorithm etc. are available with practical applications. But simultaneous search for multiple semioptimal routes are difficult with the above mentioned solutions as they produce the best one at a time. Genetic algorithm (GA) based solutions are currently available for simultaneous search of multiple routes. But the problem in finding multiple routes is that the selected routes resemble each other i.e., partly overlap. In this paper a GA based algorithm with a novel fitness function has been proposed for simultaneous search of multiple routes avoiding overlapping. Using a portion of real road map the simulation of the proposed algorithm and other currently available algorithm are done. The simulation results demonstrate the effectiveness of the proposed algorithm over other algorithms. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Edge feature extraction in digital images with the ant colony system

    Publication Year: 2004 , Page(s): 133 - 136
    Cited by:  Papers (7)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (579 KB) |  | HTML iconHTML  

    In this paper, the perceptual graph is proposed to represent the relationship between neighboring image points. The ant colony system is applied to build the perceptual graph of digital images, which makes the basis of the layered model of a machine vision system. In the experiments, the edge feature in digital images is extracted based on the proposed machine vision model. The experimental results show that the ant colony system can effectively extract image features. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Implementation method for voting of neural networks

    Publication Year: 2004 , Page(s): 98 - 101
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (560 KB) |  | HTML iconHTML  

    In this paper, four types of voting schemes generally adopted in competition neural networks has been compared in their rationality, integrity and maneuverability. A novel hardware design approach especially for the most advanced Nash voting scheme is presented. The proposed method simplifies the circuit construction by changing multiplication operations into logarithm operations. The circuit is experimentally measured with PSPICE. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A DLSI approach for content-based image classification

    Publication Year: 2004 , Page(s): 138 - 143
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (734 KB) |  | HTML iconHTML  

    Clustering images into semantically meaningful clusters using low-level visual features is a demanding and important problem in content-based image retrieval. In this paper, we investigate the feasibility of a DLSI (differential latent semantic indexing) approach in image classification. The new method applies a combined use of the projections on and the distances to the DLSI space from a differential "image" of any two images, and employs a posteriori likelihood function in measuring the similarity between an image class in the database and an image of query. Our simple experiment gives a supporting evidence of the strength of DLSI approach in capturing the intricate variability of image content contributing to a more robust context contingent classification method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fast and accurate approximation of the long wave radiation parameterization in a GCM using neural networks: evaluation of computational performance and accuracy of approximation in the NCAR CAM-2

    Publication Year: 2004 , Page(s): 57 - 62
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (751 KB) |  | HTML iconHTML  

    A new approach based on a synergetic combination of statistical/machine learning and deterministic modeling within atmospheric models is presented. The approach uses neural networks as a statistical or machine learning technique for an accurate and fast emulation or statistical approximation of model physics parameterizations. It is applied to the development of an accurate and fast approximation of an atmospheric long wave radiation parameterization for the NCAR Community Atmospheric Model, which is the most time consuming component of model physics. The developed neural network emulation is two orders of magnitude, 50-80 times, faster than the original parameterization. A comparison of the parallel 10-year climate simulations performed with the original parameterization and its neural network emulations, confirmed that these simulations produce almost identical results. The obtained results show the conceptual and practical possibility of an efficient synergetic combination of deterministic and statistical learning components within an atmospheric climate or forecast model. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Characterizing semiconductor devices using computational intelligence techniques with semiconductor automatic test system (ATE)

    Publication Year: 2004 , Page(s): 64 - 69
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (746 KB) |  | HTML iconHTML  

    Characterization of semiconductor devices is used to gather as much data about the device as possible to determine weaknesses in design or trends in the manufacturing process. This is done by varying the device specification parameters with respect to set of predefined tests and determining where the part passes or fails. The key to this process is discovering the single trip (fig. 1. pass/fail) point as accurately as possible. However, this approach cannot guarantee the robustness of device performance variation vs specification based on only a single trip point and single test analysis. This means device could still violate the specification while passing all characterization tests. In this paper, we propose a novel multiple trip point characterization concept to overcome the constraint of single trip point concept in device characterization phase. In addition, we use computational intelligence techniques to further manipulate these sets of multiple trip point values and tests based on semiconductor ATE, such that characterization trip point values with respect to different tests can be learned by neural network and fuzzy system, then performing classification task of worst case variation of device's performance vs specification. At last, the final worst-case variation can be further detected by genetic algorithm. Our experimental results demonstrate an excellent design parameter variation analysis in device characterization phase, as well as detection of a set of worst-case tests that can provoke the worst-case variation, while traditional approach was not capable of detecting them. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A functional model based on single unit recordings from Parkinsonian brain

    Publication Year: 2004 , Page(s): 30 - 34
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (563 KB) |  | HTML iconHTML  

    Artificial neuronal clusters are arranged and linearly filtered to generate signals similar to those recorded from the mid-brain regions of patients with Parkinson's disease. The goal of the research is to construct a model containing information about several aspects of recording from a neuronal cluster in-vivo. In particular, these include: number (or size) of significant neurons in the cluster, effective filtering characteristics of brain tissue between the recording electrode and each neuron, and spiking frequency of each neuron. Furthermore, models of varying size are generated based on single-unit recordings from the human brain. Results of simulations are presented and compared. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A fuzzy multiple reference model adaptive control scheme for flexible link robotic manipulator

    Publication Year: 2004 , Page(s): 162 - 167
    Cited by:  Papers (5)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (671 KB) |  | HTML iconHTML  

    In this paper a novel fuzzy logic based multiple reference model adaptive controller approach for the position control of a single link robotic manipulator is presented. The proposed fuzzy logic scheme is used for generating multiple reference models, within the model reference adaptive control (MRAC) framework, in response to changes in modes of operation or modal swings due to manipulator tip load variation. Thus the scheme is utilized to generate dynamic reference model and the overall structure is coined as fuzzy multiple reference model adaptive controller (FMRMAC). Following a rule base the fuzzy switching scheme effectively monitors changes in operating conditions due to tip load variation. A fuzzy inference engine then fires appropriate rules, which gives a fuzzified output value. Further defuzzification is performed to switch the reference model in a predefined domain. The main contribution of the paper is that the proposed approach can be performed online and is very well suitable for plants showing sudden 'jump' in operating conditions. Unlike, static multiple model algorithms for switching (noninteracting individual model-based filters) or switching dynamic algorithms (susceptible to numerical overflow), this scheme provides an interactive multiple model environment with soft switching. This approach is found to be every effective and fault tolerant. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Robustness test of CMOS circuit based on its worst case power consumption signature using ATE and GA-MIE technique

    Publication Year: 2004 , Page(s): 92 - 97
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (762 KB) |  | HTML iconHTML  

    This paper presents a diagnosis method which works with industrial semiconductor ATE for analyzing the robustness of the circuit and uses genetic algorithm (GA) with a novel multiple individuals (chromosomes) evolution (GA-MIE) technique. The term robustness in this paper refers to stability and performance of circuits with multiple sources of uncertainties. The objective is studying the worst case activity on chip based on its worst case power consumption signature with respect to a set of worst case input tests. Tests are referred to input patterns and test conditions, since the activity of CMOS circuit is a complex function of the input tests and operating parameters. For instance, the timing and voltage levels on chip can vary due to a small variation of input timing and voltage level. Traditional test and analysis approaches do not consider test condition variation. Experimental results on a test chip show the worst case active tests generated with our approach provoke the device to run slower than normal tests using typical approaches. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Segmentation of connected Arabic characters using hidden Markov models

    Publication Year: 2004 , Page(s): 115 - 119
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (659 KB)  

    Because the Arabic text is connected by nature, segmentation of Arabic text into characters is a very important task for building an Arabic OCR. Although a lot of work has been done in this area, there is no perfect technique for segmentation has been used until now. In this paper, discrete hidden Markov models are used for segmentation of Arabic words into letters. The results are very encouraging. A system has been built and used for testing the proposed algorithm and the segmentation results achieved 99%. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.