Learning kernels for support vector machines with polynomial powers of sigmoid

In the pattern recognition research field, Support Vector Machines (SVM) have been an effectiveness tool for classification purposes, being successively employed in many applications. The SVM input data is transformed into a high dimensional space using some kernel functions where linear separation is more likely. However, there are some computational drawbacks associated to SVM. One of them is the computational burden required to find out the more adequate parameters for the kernel mapping considering each non-linearly separable input data space, which reflects the performance of SVM. This paper introduces the Polynomial Powers of Sigmoid for SVM kernel mapping, and it shows their advantages over well-known kernel functions using real and synthetic datasets.

Instantaneous complex power applied to three-phase machines

The theory presented in this paper was primarily developed to give a physical interpretation for the instantaneous power flow on a three-phase induction machine, without a neutral conductor, on any operational state and may be extended to any three-phase load. It is a vectorial interpretation of the instantaneous reactive power theory presented by Akagi et al. Which, believe the authors, isn't enough developed and its physical meaning not yet completely understood. This vectorial interpretation is based on the instantaneous complex power concept defined by Torrens for single-phase, ac, steady-state circuits, and leads to a better understanding of the power phenomenon, particularly of the distortion power. This concept has been extended by the authors to three-phase systems, through the utilization of the instantaneous space vectors. The results of measurements of instantaneous complex power on a self-excited induction generator's terminals, during an over-load application transient, are presented for illustration. The compensation of reactive power proposed by Akagi is discussed and a new horizon for the theory application is opened.

Bone regeneration in surgically created deffects filled with autogenous bone: epifluorescence microscopy analysis in rats

Although the search for the ideal bone substitute has been the focus of a large number of studies, autogenous bone is still the gold standard for the filling of defects caused by pathologies and traumas, and mainly, for alveolar ridge reconstruction, allowing the titanium implants installation. OBJECTIVES: The aim of this study was to evaluate the dynamics of autogenous bone graft incorporation process to surgically created defects in rat calvaria, using epifluorescence microscopy. MATERIAL AND METHODS: Five adult male rats weighing 200-300 g were used. The animals received two 5-mm-diameter bone defects bilaterally in each parietal bone with a trephine bur under general anesthesia. Two groups of defects were formed: a control group (n=5), in which the defects were filled with blood clot, and a graft group (n=5), in which the defects were filled with autogenous bone block, removed from the contralateral defect. The fluorochromes calcein and alizarin were applied at the 7th and 30th postoperative days, respectively. The animals were killed at 35 days. RESULTS: The mineralization process was more intense in the graft group (32.09%) and occurred mainly between 7 and 30 days, the period labeled by calcein (24.66%). CONCLUSIONS: The fluorochromes showed to be appropriate to label mineralization areas. The interfacial areas between fluorochrome labels are important sources of information about the bone regeneration dynamics.

Expert system to detect incipient fauts in rotating machines using fuzzy logic and virtual instrumentation

This paper refers to the design of an expert system that captures a waveform through the use of an accelerometer, processes the signal and converts it to the frequency domain using a Fast Fourier Transformer to then, using artificial intelligence techniques, specifically Fuzzy Reasoning, it determines if there is any failure present in the underlying mode of the equipment, such as imbalance, misalignment or bearing defects.

Hybrid methods for lot sizing on parallel machines

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Model selection for discriminative restricted boltzmann machines through meta-heuristic techniques

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Controle das propriedades dinâmicas de sistemas mecânicos através da utilização de molas inteligentes

Pós-graduação em Engenharia Mecânica - FEIS

Managing Large Data Sets Using Support Vector Machines

Hundreds of Terabytes of CMS (Compact Muon Solenoid) data are being accumulated for storage day by day at the University of Nebraska-Lincoln, which is one of the eight US CMS Tier-2 sites. Managing this data includes retaining useful CMS data sets and clearing storage space for newly arriving data by deleting less useful data sets. This is an important task that is currently being done manually and it requires a large amount of time. The overall objective of this study was to develop a methodology to help identify the data sets to be deleted when there is a requirement for storage space. CMS data is stored using HDFS (Hadoop Distributed File System). HDFS logs give information regarding file access operations. Hadoop MapReduce was used to feed information in these logs to Support Vector Machines (SVMs), a machine learning algorithm applicable to classification and regression which is used in this Thesis to develop a classifier. Time elapsed in data set classification by this method is dependent on the size of the input HDFS log file since the algorithmic complexities of Hadoop MapReduce algorithms here are O(n). The SVM methodology produces a list of data sets for deletion along with their respective sizes. This methodology was also compared with a heuristic called Retention Cost which was calculated using size of the data set and the time since its last access to help decide how useful a data set is. Accuracies of both were compared by calculating the percentage of data sets predicted for deletion which were accessed at a later instance of time. Our methodology using SVMs proved to be more accurate than using the Retention Cost heuristic. This methodology could be used to solve similar problems involving other large data sets.

Capsule Locus Polymorphism among Distinct Lineages of Enterococcus faecalis Isolated from Canals of Root-filled Teeth with Periapical Lesions

Introduction: Although Enterococcus faecalis is a member of the normal microbiota, it is also a major cause of nosocomial infections. Some strains of E. faecalis produce capsule, which contributes to pathogenesis through evasion of host defenses, and its production is dependent on the capsule (cps) operon polymorphism. This study investigated cps locus polymorphism in distinct lineages of E. faecalis isolated from canals of root-filled teeth with periapical lesions. Methods: Twenty-two E. faecalis isolates were evaluated regarding the cps operon polymorphism and genetic diversity. The 3 known CPS types were determined by polymerase chain reaction. This information was correlated with multilocus sequence typing data, which were used to define genetic lineages. Results: cpsA and cpsB were the only detected genes within the cps operon in 62.5% of E. faecalis strains (14/22), indicative of genotype CPS 1, which lacks capsule expression. The essential genes in the cps operon for capsule production were detected in the remaining strains, whereas 3 belonged to genotype CPS 5 and 5 strains to genotype CPS 2. A total of 14 sequence types (STs) were resolved in 22 E. faecalis isolates. Comparison with the E. faecalis international multilocus sequence typing database revealed that 9 STs were previously found, and that the 5 STs were novel. Conclusions: Certain E. faecalis genotypes from canals of root-filled teeth with periapical lesions belong to lineages associated with capsule expression and production of multiple virulence factors, which might account for their increased pathogenic potential. (J Endod 2012;38:58-61)

Charge dynamics in half-filled Hubbard chains with finite on-site interaction

We study the charge dynamic structure factor of the one-dimensional Hubbard model with finite on-site repulsion U at half-filling. Numerical results from the time-dependent density matrix renormalization group are analyzed by comparison with the exact spectrum of the model. The evolution of the line shape as a function of U is explained in terms of a relative transfer of spectral weight between the two-holon continuum that dominates in the limit U -> infinity and a subset of the two-holon-two-spinon continuum that reconstructs the electron-hole continuum in the limit U -> 0. Power-law singularities along boundary lines of the spectrum are described by effective impurity models that are explicitly invariant under spin and eta-spin SU(2) rotations. The Mott-Hubbard metal-insulator transition is reflected in a discontinuous change of the exponents of edge singularities at U = 0. The sharp feature observed in the spectrum for momenta near the zone boundary is attributed to a van Hove singularity that persists as a consequence of integrability.

Electromagnetic torque ripple and copper losses reduction in permanent magnet synchronous machines

This paper presents a method for electromagnetic torque ripple and copper losses reduction in (non-sinusoidal or trapezoidal) surface-mount permanent magnet synchronous machines (SM-PMSM). The method is based on an extension of classical dq transformation that makes it possible to write a vectorial model for this kind of machine (with a non-sinusoidal back-EMF waveform). This model is obtained by the application of that transformation in the classical machine per-phase model. That transformation can be applied to machines that have any type of back-EMF waveform, and not only trapezoidal or square-wave back-EMF waveforms. Implementation results are shown for an electrical converter, using the proposed vectorial model, feeding a non-sinusoidal synchronous machine (brushless DC motor). They show that the use of this vectorial mode is a way to achieve improvements in the performance of this kind of machine, considering the electromagnetic torque ripple and copper losses, if compared to a drive system that employs a classical six-step mode as a converter. Copyright (C) 2011 John Wiley & Sons, Ltd.

Combining meta-learning and search techniques to select parameters for support vector machines

Support Vector Machines (SVMs) have achieved very good performance on different learning problems. However, the success of SVMs depends on the adequate choice of the values of a number of parameters (e.g., the kernel and regularization parameters). In the current work, we propose the combination of meta-learning and search algorithms to deal with the problem of SVM parameter selection. In this combination, given a new problem to be solved, meta-learning is employed to recommend SVM parameter values based on parameter configurations that have been successfully adopted in previous similar problems. The parameter values returned by meta-learning are then used as initial search points by a search technique, which will further explore the parameter space. In this proposal, we envisioned that the initial solutions provided by meta-learning are located in good regions of the search space (i.e. they are closer to optimum solutions). Hence, the search algorithm would need to evaluate a lower number of candidate solutions when looking for an adequate solution. In this work, we investigate the combination of meta-learning with two search algorithms: Particle Swarm Optimization and Tabu Search. The implemented hybrid algorithms were used to select the values of two SVM parameters in the regression domain. These combinations were compared with the use of the search algorithms without meta-learning. The experimental results on a set of 40 regression problems showed that, on average, the proposed hybrid methods obtained lower error rates when compared to their components applied in isolation.

Fault detection in rotating machines by vibration signal processing techniques

Machines with moving parts give rise to vibrations and consequently noise. The setting up and the status of each machine yield to a peculiar vibration signature. Therefore, a change in the vibration signature, due to a change in the machine state, can be used to detect incipient defects before they become critical. This is the goal of condition monitoring, in which the informations obtained from a machine signature are used in order to detect faults at an early stage. There are a large number of signal processing techniques that can be used in order to extract interesting information from a measured vibration signal. This study seeks to detect rotating machine defects using a range of techniques including synchronous time averaging, Hilbert transform-based demodulation, continuous wavelet transform, Wigner-Ville distribution and spectral correlation density function. The detection and the diagnostic capability of these techniques are discussed and compared on the basis of experimental results concerning gear tooth faults, i.e. fatigue crack at the tooth root and tooth spalls of different sizes, as well as assembly faults in diesel engine. Moreover, the sensitivity to fault severity is assessed by the application of these signal processing techniques to gear tooth faults of different sizes.