Tracking the states of a nonlinear and nonstationary system in the weight-space of artificial neural networks

We propose a novel interpretation and usage of Neural Network (NN) in modeling physiological signals, which are allowed to be nonlinear and/or nonstationary. The method consists of training a NN for the k-step prediction of a physiological signal, and then examining the connection-weight-space (CWS) of the NN to extract information about the signal generator mechanism. We de. ne a novel feature, Normalized Vector Separation (gamma(ij)), to measure the separation of two arbitrary states i and j in the CWS and use it to track the state changes of the generating system. The performance of the method is examined via synthetic signals and clinical EEG. Synthetic data indicates that gamma(ij) can track the system down to a SNR of 3.5 dB. Clinical data obtained from three patients undergoing carotid endarterectomy of the brain showed that EEG could be modeled (within a root-means-squared-error of 0.01) by the proposed method, and the blood perfusion state of the brain could be monitored via gamma(ij), with small NNs having no more than 21 connection weight altogether.

Two Bayesian treatments of the n-tuple recognition method

Two probabilistic interpretations of the n-tuple recognition method are put forward in order to allow this technique to be analysed with the same Bayesian methods used in connection with other neural network models. Elementary demonstrations are then given of the use of maximum likelihood and maximum entropy methods for tuning the model parameters and assisting their interpretation. One of the models can be used to illustrate the significance of overlapping n-tuple samples with respect to correlations in the patterns.

Novel imprinted polymers as artificial enzymes

Derivatives of L-histidine were investigated as suitable models for the Asp-His couple found in the catalytic triad of serine proteases. A combination of molecular dynamics and IH NMR spectroscopy suggested that the most populous conformations of N-acetyl-L-histidine and the N-acetyl-L-histidine anion were predominated by those in which the carboxylate group was gauche to the imidazole ring overcoming steric and electrostatic repulsion, suggesting there is an interaction between the carboxylate group and the imidazole ring. Kinetic studies, using imidazole, N-acetyl-L-histidine and the N-acetyl-L-histidine anion showed that in a DMSO/H20 9: 1 v/v solution, the N-acetyl-L-histidine anion catalysed the hydrolysis of p-nitrophenyl acetate at a greater rate than using either imidazole or N-acetyl-L-histidine as catalyst. This indicates that the carboxylate group affects the nucleophilicity of the unprotonated imidazole ring. 31P MAS NMR spectroscopy was investigated as a new technique for the study of the template molecule environment within the polymer networks. It was found that it was possible to distinguish between template associated with the polymer and that which was precipitated onto the surface, though it was not possible to distinguish between polymer within imprinted cavities and that which was not. Attempts to study the effect of the carboxylate group/imidazole ring interaction in the imprinted cavity of a molecularly imprinted polymer network were hindered by the method used to follow the reaction. It was found though that in a pH 8.0 buffered solution the presence of imprinted cavities increased the rate of reaction for those polymers derived from L-histidine. Some preliminary investigations into the design and synthesis of an MIP which would catalyse the oxy-Cope rearrangement were carried out but the results were inconclusive.

A biological-inspired support frame for an artificial cornea

Bilateral corneal blindness represents a quarter of the total blind, world-wide. The artificial cornea in assorted forms, was developed to replace opaque non-functional corneas and to return sight in otherwise hopeless cases that were not amenable to corneal grafts; believed to be 2% of corneal blind. Despite technological advances in materials design and tissue engineering no artificial cornea has provided absolute, long-term success. Formidable problems exist, due to a combination of unpredictable wound healing and unmanageable pathology. To have a solid guarantee of reliable success an artificial cornea must possess three attributes: an optical window to replace the opaque cornea; a strong, long term union to surrounding ocular tissue; and the ability to induce desired host responses. A unique artificial cornea possesses all three functional attributes- the Osteo-odonto-keratoprosthesis (OOKP). The OOKP has a high success rate and can survive for up to twenty years, but it is complicated both in structure and in surgical procedure; it is expensive and not universally available. The aim of this project was to develop a synthetic substitute for the OOKP, based upon key features of the tooth and bone structure. In doing so, surgical complexity and biological complications would be reduced. Analysis of the biological effectiveness of the OOKP showed that the structure of bone was the most crucial component for implant retention. An experimental semi-rigid hydroxyapatite framework was fabricated with a complex bone-like architecture, which could be fused to the optical window. The first method for making such a framework, was pressing and sintering of hydroxyapatite powders; however, it was not possible to fabricate a void architecture with the correct sizes and uniformity of pores. Ceramers were synthesised using alternative pore forming methods, providing for improved mechanical properties and stronger attachment to the plastic optical window. Naturally occurring skeletal structures closely match the structural features of all forms of natural bone. Synthetic casts were fabricated using the replamineform process, of desirable natural artifacts, such as coral and sponges. The final method of construction by-passed ceramic fabrication in favour of pre-formed coral derivatives and focused on methods for polymer infiltration, adhesion and fabrication. Prototypes were constructed and evaluated; a fully penetrative synthetic OOKP analogue was fabricated according to the dimensions of the OOKP. Fabrication of the cornea shaped OOKP synthetic analogue was also attempted.

Evaluating the performance of a EuroDivisia index using artificial intelligence techniques

We compare two methods in order to predict inflation rates in Europe. One method uses a standard back propagation neural network and the other uses an evolutionary approach, where the network weights and the network architecture is evolved. Results indicate that back propagation produces superior results. However, the evolving network still produces reasonable results with the advantage that the experimental set-up is minimal. Also of interest is the fact that the Divisia measure of money is superior as a predictive tool over simple sum.

Evaluating the performance of a EuroDivisia index using artificial intelligence techniques

This paper compares two methods to predict in°ation rates in Europe. One method uses a standard back propagation neural network and the other uses an evolutionary approach, where the network weights and the network architecture are evolved. Results indicate that back propagation produces superior results. However, the evolving network still produces reasonable results with the advantage that the experimental set-up is minimal. Also of interest is the fact that the Divisia measure of money is superior as a predictive tool over simple sum.

Evaluation efficiency of large-scale data set with negative data:an artificial neural network approach

Data envelopment analysis (DEA) is the most widely used methods for measuring the efficiency and productivity of decision-making units (DMUs). The need for huge computer resources in terms of memory and CPU time in DEA is inevitable for a large-scale data set, especially with negative measures. In recent years, wide ranges of studies have been conducted in the area of artificial neural network and DEA combined methods. In this study, a supervised feed-forward neural network is proposed to evaluate the efficiency and productivity of large-scale data sets with negative values in contrast to the corresponding DEA method. Results indicate that the proposed network has some computational advantages over the corresponding DEA models; therefore, it can be considered as a useful tool for measuring the efficiency of DMUs with (large-scale) negative data.

Laser-induced breakdown spectroscopy quantitative analysis method via adaptive analytical line selection and relevance vector machine regression model

Abstract A new LIBS quantitative analysis method based on analytical line adaptive selection and Relevance Vector Machine (RVM) regression model is proposed. First, a scheme of adaptively selecting analytical line is put forward in order to overcome the drawback of high dependency on a priori knowledge. The candidate analytical lines are automatically selected based on the built-in characteristics of spectral lines, such as spectral intensity, wavelength and width at half height. The analytical lines which will be used as input variables of regression model are determined adaptively according to the samples for both training and testing. Second, an LIBS quantitative analysis method based on RVM is presented. The intensities of analytical lines and the elemental concentrations of certified standard samples are used to train the RVM regression model. The predicted elemental concentration analysis results will be given with a form of confidence interval of probabilistic distribution, which is helpful for evaluating the uncertainness contained in the measured spectra. Chromium concentration analysis experiments of 23 certified standard high-alloy steel samples have been carried out. The multiple correlation coefficient of the prediction was up to 98.85%, and the average relative error of the prediction was 4.01%. The experiment results showed that the proposed LIBS quantitative analysis method achieved better prediction accuracy and better modeling robustness compared with the methods based on partial least squares regression, artificial neural network and standard support vector machine.

Estimation of average annual daily traffic (AADT) using geographically weighted regression (GWR) method and geographic information system (GIS)

Annual average daily traffic (AADT) is important information for many transportation planning, design, operation, and maintenance activities, as well as for the allocation of highway funds. Many studies have attempted AADT estimation using factor approach, regression analysis, time series, and artificial neural networks. However, these methods are unable to account for spatially variable influence of independent variables on the dependent variable even though it is well known that to many transportation problems, including AADT estimation, spatial context is important. ^ In this study, applications of geographically weighted regression (GWR) methods to estimating AADT were investigated. The GWR based methods considered the influence of correlations among the variables over space and the spatially non-stationarity of the variables. A GWR model allows different relationships between the dependent and independent variables to exist at different points in space. In other words, model parameters vary from location to location and the locally linear regression parameters at a point are affected more by observations near that point than observations further away. ^ The study area was Broward County, Florida. Broward County lies on the Atlantic coast between Palm Beach and Miami-Dade counties. In this study, a total of 67 variables were considered as potential AADT predictors, and six variables (lanes, speed, regional accessibility, direct access, density of roadway length, and density of seasonal household) were selected to develop the models. ^ To investigate the predictive powers of various AADT predictors over the space, the statistics including local r-square, local parameter estimates, and local errors were examined and mapped. The local variations in relationships among parameters were investigated, measured, and mapped to assess the usefulness of GWR methods. ^ The results indicated that the GWR models were able to better explain the variation in the data and to predict AADT with smaller errors than the ordinary linear regression models for the same dataset. Additionally, GWR was able to model the spatial non-stationarity in the data, i.e., the spatially varying relationship between AADT and predictors, which cannot be modeled in ordinary linear regression. ^

A static traffic assignment model combined with an artificial neural network delay model

As traffic congestion continues to worsen in large urban areas, solutions are urgently sought. However, transportation planning models, which estimate traffic volumes on transportation network links, are often unable to realistically consider travel time delays at intersections. Introducing signal controls in models often result in significant and unstable changes in network attributes, which, in turn, leads to instability of models. Ignoring the effect of delays at intersections makes the model output inaccurate and unable to predict travel time. To represent traffic conditions in a network more accurately, planning models should be capable of arriving at a network solution based on travel costs that are consistent with the intersection delays due to signal controls. This research attempts to achieve this goal by optimizing signal controls and estimating intersection delays accordingly, which are then used in traffic assignment. Simultaneous optimization of traffic routing and signal controls has not been accomplished in real-world applications of traffic assignment. To this end, a delay model dealing with five major types of intersections has been developed using artificial neural networks (ANNs). An ANN architecture consists of interconnecting artificial neurons. The architecture may either be used to gain an understanding of biological neural networks, or for solving artificial intelligence problems without necessarily creating a model of a real biological system. The ANN delay model has been trained using extensive simulations based on TRANSYT-7F signal optimizations. The delay estimates by the ANN delay model have percentage root-mean-squared errors (%RMSE) that are less than 25.6%, which is satisfactory for planning purposes. Larger prediction errors are typically associated with severely oversaturated conditions. A combined system has also been developed that includes the artificial neural network (ANN) delay estimating model and a user-equilibrium (UE) traffic assignment model. The combined system employs the Frank-Wolfe method to achieve a convergent solution. Because the ANN delay model provides no derivatives of the delay function, a Mesh Adaptive Direct Search (MADS) method is applied to assist in and expedite the iterative process of the Frank-Wolfe method. The performance of the combined system confirms that the convergence of the solution is achieved, although the global optimum may not be guaranteed.

In vivo biocompatibility evaluation of composite polymeric materials for use in a novel artificial heart valve

A novel trileaflet polymer valve is a composite design of a biostable polymer poly(styrene-isobutylene-styrene) (SIBS) with a reinforcement polyethylene terephthalate (PET) fabric. Surface roughness and hydrophilicity vary with fabrication methods and influence leaflet biocompatibility. The purpose of this study was to investigate the biocompatibility of this composite material using both small animal (nonfunctional mode) and large animal (functional mode) models. Composite samples were manufactured using dip coating and solvent casting with different coating thickness (251μm and 50μm). Sample's surface was characterized through qualitative SEM observation and quantitative surface roughness analysis. A novel rat abdominal aorta model was developed to test the composite samples in a similar pulsatile flow condition as its intended use. The sample's tissue response was characterized by histological examination. Among the samples tested, the 25μm solvent-cast sample exhibited the smoothest surface and best biocompatibility in terms of tissue capsulation thickness, and was chosen as the method for fabrication of the SIBS valve. Phosphocholine was used to create a hydrophilic surface on selected composite samples, which resulted in improved blood compatibility. Four SIBS valves (two with phosphocholine modification) were implanted into sheep. Echocardiography, blood chemistry, and system pathology were conducted to evaluate the valve's performance and biocompatibility. No adverse response was identified following implantation. The average survival time was 76 days, and one sheep with the phosphocholine modified valve passed the FDA minimum requirement of 140 days with approximately 20 million cycles of valve activity. The explanted valves were observed under the aid of a dissection microscope, and evaluated via histology, SEM and X-ray. Surface cracks and calcified tissue deposition were found on the leaflets. In conclusion, we demonstrated the applicability of using a new rat abdominal aorta model for biocompatibility assessment of polymeric materials. A smooth and complete coating surface is essential for the biocompatibility of PET/SIBS composite, and surface modification using phosphocholine improves blood compatibility. Extrinsic calcification was identified on the leaflets and was associated with regions of surface cracks.

A research platform for artificial neural networks with applications in pediatric epilepsy

This dissertation established a state-of-the-art programming tool for designing and training artificial neural networks (ANNs) and showed its applicability to brain research. The developed tool, called NeuralStudio, allows users without programming skills to conduct studies based on ANNs in a powerful and very user friendly interface. A series of unique features has been implemented in NeuralStudio, such as ROC analysis, cross-validation, network averaging, topology optimization, and optimization of the activation function’s slopes. It also included a Support Vector Machines module for comparison purposes. Once the tool was fully developed, it was applied to two studies in brain research. In the first study, the goal was to create and train an ANN to detect epileptic seizures from subdural EEG. This analysis involved extracting features from the spectral power in the gamma frequencies. In the second application, a unique method was devised to link EEG recordings to epileptic and nonepileptic subjects. The contribution of this method consisted of developing a descriptor matrix that can be used to represent any EEG file regarding its duration and the number of electrodes. The first study showed that the inter-electrode mean of the spectral power in the gamma frequencies and its duration above a specific threshold performs better than the other frequencies in seizure detection, exhibiting an accuracy of 95.90%, a sensitivity of 92.59%, and a specificity of 96.84%. The second study yielded that Hjorth’s parameter activity is sufficient to accurately relate EEG to epileptic and non-epileptic subjects. After testing, accuracy, sensitivity and specificity of the classifier were all above 0.9667. Statistical tests measured the superiority of activity at over 99.99 % certainty. It was demonstrated that (1) the spectral power in the gamma frequencies is highly effective in locating seizures from EEG and (2) activity can be used to link EEG recordings to epileptic and non-epileptic subjects. These two studies required high computational load and could be addressed thanks to NeuralStudio. From a medical perspective, both methods proved the merits of NeuralStudio in brain research applications. For its outstanding features, NeuralStudio has been recently awarded a patent (US patent No. 7502763).

A multi-method approach for the assessment of composite indices and rankings

There is growing popularity in the use of composite indices and rankings for cross-organizational benchmarking. However, little attention has been paid to alternative methods and procedures for the computation of these indices and how the use of such methods may impact the resulting indices and rankings. This dissertation developed an approach for assessing composite indices and rankings based on the integration of a number of methods for aggregation, data transformation and attribute weighting involved in their computation. The integrated model developed is based on the simulation of composite indices using methods and procedures proposed in the area of multi-criteria decision making (MCDM) and knowledge discovery in databases (KDD). The approach developed in this dissertation was automated through an IT artifact that was designed, developed and evaluated based on the framework and guidelines of the design science paradigm of information systems research. This artifact dynamically generates multiple versions of indices and rankings by considering different methodological scenarios according to user specified parameters. The computerized implementation was done in Visual Basic for Excel 2007. Using different performance measures, the artifact produces a number of excel outputs for the comparison and assessment of the indices and rankings. In order to evaluate the efficacy of the artifact and its underlying approach, a full empirical analysis was conducted using the World Bank's Doing Business database for the year 2010, which includes ten sub-indices (each corresponding to different areas of the business environment and regulation) for 183 countries. The output results, which were obtained using 115 methodological scenarios for the assessment of this index and its ten sub-indices, indicated that the variability of the component indicators considered in each case influenced the sensitivity of the rankings to the methodological choices. Overall, the results of our multi-method assessment were consistent with the World Bank rankings except in cases where the indices involved cost indicators measured in per capita income which yielded more sensitive results. Low income level countries exhibited more sensitivity in their rankings and less agreement between the benchmark rankings and our multi-method based rankings than higher income country groups.

Online solid phase extraction liquid chromatography tandem mass spectrometry (SPE-LC-MS/MS) method for the determination of sucralose in reclaimed and drinking waters and its photo degradation in natural waters from South Florida

Background Sucralose has gained popularity as a low calorie artificial sweetener worldwide. Due to its high stability and persistence, sucralose has shown widespread occurrence in environmental waters, at concentrations that could reach up to several μg/L. Previous studies have used time consuming sample preparation methods (offline solid phase extraction/derivatization) or methods with rather high detection limits (direct injection) for sucralose analysis. This study described a faster and sensitive analytical method for the determination of sucralose in environmental samples. Results An online SPE-LC–MS/MS method was developed, being capable to quantify sucralose in 12 minutes using only 10 mL of sample, with method detection limits (MDLs) of 4.5 ng/L, 8.5 ng/L and 45 ng/L for deionized water, drinking and reclaimed waters (1:10 diluted with deionized water), respectively. Sucralose was detected in 82% of the reclaimed water samples at concentrations reaching up to 18 μg/L. The monthly average for a period of one year was 9.1 ± 2.9 μg/L. The calculated mass loads per capita of sucralose discharged through WWTP effluents based on the concentrations detected in wastewaters in the U. S. is 5.0 mg/day/person. As expected, the concentrations observed in drinking water were much lower but still relevant reaching as high as 465 ng/L. In order to evaluate the stability of sucralose, photodegradation experiments were performed in natural waters. Significant photodegradation of sucralose was observed only in freshwater at 254 nm. Minimal degradation (<20%) was observed for all matrices under more natural conditions (350 nm or solar simulator). The only photolysis product of sucralose identified by high resolution mass spectrometry was a de-chlorinated molecule at m/z 362.0535, with molecular formula C12H20Cl2O8. Conclusions Online SPE LC-APCI/MS/MS developed in the study was applied to more than 100 environmental samples. Sucralose was frequently detected (>80%) indicating that the conventional treatment process employed in the sewage treatment plants is not efficient for its removal. Detection of sucralose in drinking waters suggests potential contamination of surface and ground waters sources with anthropogenic wastewater streams. Its high resistance to photodegradation, minimal sorption and high solubility indicate that sucralose could be a good tracer of anthropogenic wastewater intrusion into the environment.

A Static Traffic Assignment Model Combined with an Artificial Neural Network Delay Model

As traffic congestion continues to worsen in large urban areas, solutions are urgently sought. However, transportation planning models, which estimate traffic volumes on transportation network links, are often unable to realistically consider travel time delays at intersections. Introducing signal controls in models often result in significant and unstable changes in network attributes, which, in turn, leads to instability of models. Ignoring the effect of delays at intersections makes the model output inaccurate and unable to predict travel time. To represent traffic conditions in a network more accurately, planning models should be capable of arriving at a network solution based on travel costs that are consistent with the intersection delays due to signal controls. This research attempts to achieve this goal by optimizing signal controls and estimating intersection delays accordingly, which are then used in traffic assignment. Simultaneous optimization of traffic routing and signal controls has not been accomplished in real-world applications of traffic assignment. To this end, a delay model dealing with five major types of intersections has been developed using artificial neural networks (ANNs). An ANN architecture consists of interconnecting artificial neurons. The architecture may either be used to gain an understanding of biological neural networks, or for solving artificial intelligence problems without necessarily creating a model of a real biological system. The ANN delay model has been trained using extensive simulations based on TRANSYT-7F signal optimizations. The delay estimates by the ANN delay model have percentage root-mean-squared errors (%RMSE) that are less than 25.6%, which is satisfactory for planning purposes. Larger prediction errors are typically associated with severely oversaturated conditions. A combined system has also been developed that includes the artificial neural network (ANN) delay estimating model and a user-equilibrium (UE) traffic assignment model. The combined system employs the Frank-Wolfe method to achieve a convergent solution. Because the ANN delay model provides no derivatives of the delay function, a Mesh Adaptive Direct Search (MADS) method is applied to assist in and expedite the iterative process of the Frank-Wolfe method. The performance of the combined system confirms that the convergence of the solution is achieved, although the global optimum may not be guaranteed.