Multiclass Support Vector Machines for Classification of ECG Data with Missing Values

The article presents an experimental study on multiclass Support Vector Machine (SVM) methods over a cardiac arrhythmia dataset that has missing attribute values for electrocardiogram (ECG) diagnostic application. The presence of an incomplete dataset and high data dimensionality can affect the performance of classifiers. Imputation of missing data and discriminant analysis are commonly used as preprocessing techniques in such large datasets. The article proposes experiments to evaluate performance of One-Against-All (OAA) and One-Against-One (OAO) approaches in kernel multiclass SVM for a heartbeat classification problem with imputation and dimension reduction techniques. The results indicate that the OAA approach has superiority over OAO in multiclass SVM for ECG data analysis with missing values.     

Measures of localized spinal muscle fatigue

The objective of this study was to determine the best variable, if any, to indicate the level of localized muscle fatigue. Six male and six female subjects were studied while they exerted their maximal voluntary contraction and 40% of maximal voluntary contraction of spinal extensors in an isometric lifting activity. The electromyography (EMG) of erectores spinae at thoracic and lumbar levels was measured bilaterally. Also, the muscle bed blood volume, level of blood oxygenation to erector spinae at L3 level and heart rate were measured. The initial and final values of subjective feelings of fatigue through visual analogue score, rate of perceived exertions and body part discomfort rating were recorded. The mean maximal voluntary contraction forces for males and females were 899 (238) N and 431 (135) N respectively. The mean durations of hold in maximal voluntary fatiguing contraction were 68.1 (39.9) s and 72.3 (37.0) s for men and women respectively. By the end of the hold the force declined to 52% for males and 62% for females. The EMG amplitudes and median frequencies also progressively declined (p < 0.01). ANOVA revealed that the task percentile values of all variables were significantly different (p < 0.01). Gender had a significant main effect (p < 0.01). The correlation coefficients between force and other individual variables were weak to modest, but significant (p < 0.001). None of the single variables predicted fatigue reliably for either gender and levels of contraction. The regression equations developed were highly significant (p < 0.01) and they explained 96 to 98% of variance in both genders and contractions.     

Enhancement of speech signals separated from their convolutive mixture by FDICA algorithm

This paper presents a novel method for the enhancement of independent components of mixed speech signal segregated by the frequency domain independent component analysis (FDICA) algorithm. The enhancement algorithm proposed here is based on maximum a posteriori (MAP) estimation of the speech spectral components using generalized Gaussian distribution (GGD) function as the statistical model for the time–frequency series of speech (TFSS) signal. The proposed MAP estimator has been used and evaluated as the post-processing stage for the separation of convolutive mixture of speech signals by the fixed-point FDICA algorithm. It has been found that the combination of separation algorithm with the proposed enhancement algorithm provides better separation performance under both the reverberant and non-reverberant conditions.     

Reasoning about Qualitative Spatial Relationships

In this paper, we consider various spatial relationships that are of general interest in pictorial database systems and other applications. We present a set of rules that allow us to deduce new relationships from a given set of relationships. A deductive mechanism using these rules can be used in query-processing systems that retrieve pictures by content. The given set of rules is shown to be sound; that is, the deductions are logically correct. The rules are also shown to be complete for three-dimensional systems; that is, every relationship that is implied by a given consistent set of relationships F is deducible from F using the given rules. In addition, we show that the given set of rules is incomplete for two-dimensional systems. We also present efficient algorithms for the deduction and reduction problems. The deduction problem consists of computing all the relationships deducible from a given set, while the reduction problem consists of computing a minimal subset of a given set of relationships that implies all the relationships in the given set.     

Modelling IP Mobility

We study a highly simplified version of the proposed mobility support in version 6 of Internet Protocols (IP). We concentrate on the issue of ensuring that messages to and from mobile agents are delivered without loss of connectivity. We provide three models, of increasingly complex nature, of a network of routers and computing agents that are interconnected via the routers: the first is without mobile agents and is treated as a specification for the next two; the second supports mobile agents, and the third additionally allows correspondent agents to cache the current location of a mobile agent. Following a detailed analysis of the three models to extract invariant properties, we show that the three models are related by a suitable notion of equivalence based on barbed bisimulation. Finally, we report on some experiments in simulating and verifying finite state versions of our model.     

RVGEN: a tool for generation of random variates

A number of applications in simulation and software testing require random number generation, both uniform and non-uniform. Although libraries are available for random number generation, there is no user-friendly tool to enable the user to use and build random number generators. This paper presents the RVGEN tool, developed at TRDDC. Using the tool, the software developer can design a random number generator specific to a particular pattern. The tool takes specifications at a high level and also partly in C++, and generates C++ code for a number of user-friendly functions. These include setting and getting of random variate parameters, GUIs for inputting parameters, validation of parameters, parameter input and output from a file, generation of random variate class declarations, generation of seeds for multiple streams, and testing of the random variate generator. The tool provides a number of classical, general-purpose and empiric distributions by default.     

A discrete event approach to the command sequence replanning and control in a teleprogramming system

A command sequence replanning and control method, which enables the slave system to autonomously recover from error conditions, is proposed in a telerobot system. A task model described as a form of the controlled Petri net (CPN) is used as a prior knowledge for the slave system to carry out the given task successfully without the operator's aid even in unexpected error conditions. The CPN model incorporates the contact states and transitional motions between them that possibly exist in the task execution process. The motion command is automatically generated from the master system and transmitted to the slave whenever the contact state changes in the master model. Referring to the CPN model and the given motion commands, the slave system detects if the actual contact state is unexpected and then plans the recovery path from the unexpected state. The feasibility of the command sequence replanning and control algorithm is verified through an example to perform a simple part‐mating task. © 2001 John Wiley & Sons, Inc.     

Mammalian abasic site base excision repair. Identification of the reaction sequence and rate-determining steps

Base excision repair (BER) is one of the cellular defense mechanisms repairing damage to nucleoside 5'-monophosphate residues in genomic DNA. This repair pathway is initiated by spontaneous or enzymatic N-glycosidic bond cleavage creating an abasic or apurinic-apyrimidinic (AP) site in double-stranded DNA. Class II AP endonuclease, deoxyribonucleotide phosphate (dRP) lyase, DNA synthesis, and DNA ligase activities complete repair of the AP site. In mammalian cell nuclear extract, BER can be mediated by a macromolecular complex containing DNA polymerase beta (beta-pol) and DNA ligase I. These two enzymes are capable of contributing the latter three of the four BER enzymatic activities. In the present study, we found that AP site BER can be reconstituted in vitro using the following purified human proteins: AP endonuclease, beta-pol, and DNA ligase I. Examination of the individual enzymatic steps in BER allowed us to identify an ordered reaction pathway: subsequent to 5' "nicking" of the AP site-containing DNA strand by AP endonuclease, beta-pol performs DNA synthesis prior to removal of the 5'-dRP moiety in the gap. Removal of the dRP flap is strictly required for DNA ligase I to seal the resulting nick. Additionally, the catalytic rate of the reconstituted BER system and the individual enzymatic activities was measured. The reconstituted BER system performs repair of AP site DNA at a rate that is slower than the respective rates of AP endonuclease, DNA synthesis, and ligation, suggesting that these steps are not rate-determining in the overall reconstituted BER system. Instead, the rate-limiting step in the reconstituted system was found to be removal of dRP (i.e. dRP lyase), catalyzed by the amino-terminal domain of beta-pol. This work is the first to measure the rate of BER in an in vitro reaction. The potential significance of the dRP-containing intermediate in the regulation of BER is discussed.     

Grain size effects on the dielectric properties of ferroelectric Bi2VO5.5 ceramics

Dielectric properties and microstructural characteristics of ferroelectric bismuth vanadate (Bi₂VO_5.5) ceramics exhibiting grain sizes of 7, 10, 20 and 25 m have been studied. Microstructural studies indicate the presence of ferroelectric 90° domain patterns on the surface as well as in the bulk of the coarse-grained ceramics. The dielectric constant and the loss tangent both at room temperature and in the vicinity of the Curie temperature have been found to increase with increasing grain size. The Curie temperature (725 K) is found to shift slightly (by about 7 K) towards higher temperatures as the grain size increases (7–25 m). The magnitude of the dielectric anomaly around 725 K is found to be higher for coarse-grained ceramics. The dielectric constant and the loss have been found to decrease with increase in frequency (1–100 kHz) for all the ceramics studied. The increase in dielectric constant with increasing grain size is attributed to a decrease in thickness of the relatively more insulating grain boundary layer.     

Impedance analysis of Pb<Subscript>2</Subscript>Sb<Subscript>3</Subscript>LaTi<Subscript>5</Subscript>O<Subscript>18</Subscript> ceramic

Polycrystalline sample of Pb₂Sb₃LaTi₅O₁₈, a member of tungsten- bronze (TB family, was prepared using a high temperature solid- state reaction technique. XRD analysis indicated the formation of a single-phase orthorhombic structure. The dielectric studies revealed the diffuse phase transition and the transition temperature was found to be at 52° C. Impedance plots were used as tools to analyse the sample behaviour as a function of frequency. Cole-Cole plots showed Debye relaxation. The activation energy was estimated to be 0·634 eV from the temperature variation of d.c. conductivity. The nature of variation of d.c. conductivity with temperature suggested NTCR behaviour.