Pruning methods to MLP neural networks considering proportional apparent error rate for classification problems with unbalanced data

This article deals with classification problems involving unequal probabilities in each class and discusses metrics to systems that use multilayer perceptrons neural networks (MLP) for the task of classifying new patterns. In addition we propose three new pruning methods that were compared to other seven existing methods in the literature for MLP networks. All pruning algorithms presented in this paper have been modified by the authors to do pruning of neurons, in order to produce fully connected MLP networks but being small in its intermediary layer. Experiments were carried out involving the E. coli unbalanced classification problem and ten pruning methods. The proposed methods had obtained good results, actually, better results than another pruning methods previously defined at the MLP neural network area. (C) 2014 Elsevier Ltd. All rights reserved.

Spectrum Distribution in Cognitive Radio: Error Correcting Codes Perspective

Cognitive radio is a growing zone in wireless communication which offers an opening in complete utilization of incompetently used frequency spectrum: deprived of crafting interference for the primary (authorized) user, the secondary user is indorsed to use the frequency band. Though, scheming a model with the least interference produced by the secondary user for primary user is a perplexing job. In this study we proposed a transmission model based on error correcting codes dealing with a countable number of pairs of primary and secondary users. However, we obtain an effective utilization of spectrum by the transmission of the pairs of primary and secondary users' data through the linear codes with different given lengths. Due to the techniques of error correcting codes we developed a number of schemes regarding an appropriate bandwidth distribution in cognitive radio.

Fourier series for quaternions and the square of the error theorem

In this paper we introduce a type of Hypercomplex Fourier Series based on Quaternions, and discuss on a Hypercomplex version of the Square of the Error Theorem. Since their discovery by Hamilton (Sinegre [1]), quaternions have provided beautifully insights either on the structure of different areas of Mathematics or in the connections of Mathematics with other fields. For instance: I) Pauli spin matrices used in Physics can be easily explained through quaternions analysis (Lan [2]); II) Fundamental theorem of Algebra (Eilenberg [3]), which asserts that the polynomial analysis in quaternions maps into itself the four dimensional sphere of all real quaternions, with the point infinity added, and the degree of this map is n. Motivated on earlier works by two of us on Power Series (Pendeza et al. [4]), and in a recent paper on Liouville’s Theorem (Borges and Mar˜o [5]), we obtain an Hypercomplex version of the Fourier Series, which hopefully can be used for the treatment of hypergeometric partial differential equations such as the dumped harmonic oscillation.

Square of the Error Octonionic Theorem and Hypercomplex Fourier Series

The focus of this paper is to address some classical results for a class of hypercomplex numbers. More specifically we present an extension of the Square of the Error Theorem and a Bessel inequality for octonions.

A decoding procedure which improves code rate and error corrections

Corresponding to $C_{0}[n,n-r]$, a binary cyclic code generated by a primitive irreducible polynomial $p(X)\in \mathbb{F}_{2}[X]$ of degree $r=2b$, where $b\in \mathbb{Z}^{+}$, we can constitute a binary cyclic code $C[(n+1)^{3^{k}}-1,(n+1)^{3^{k}}-1-3^{k}r]$, which is generated by primitive irreducible generalized polynomial $p(X^{\frac{1}{3^{k}}})\in \mathbb{F}_{2}[X;\frac{1}{3^{k}}\mathbb{Z}_{0}]$ with degree $3^{k}r$, where $k\in \mathbb{Z}^{+}$. This new code $C$ improves the code rate and has error corrections capability higher than $C_{0}$. The purpose of this study is to establish a decoding procedure for $C_{0}$ by using $C$ in such a way that one can obtain an improved code rate and error-correcting capabilities for $C_{0}$.

Escala de estresse percebido aplicada a estudantes universitárias: estudo de validação

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

Preocupação com a forma do corpo de graduandos de Farmácia-Bioquímica

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Specific determination of maximal lactate steady state in soccer players

The aim of this study was to establish the validity of the anaerobic threshold (AT) determined on the soccer-specific Hoff circuit (AT(Hoff)) to predict the maximal lactate steady-state exercise intensity (MLSSHoff) with the ball. Sixteen soccer players (age: 16.0 +/- 0.5 years; body mass: 63.7 +/- 9.0 kg; and height: 169.4 +/- 5.3 cm) were submitted to 5 progressive efforts (7.0-11.0 km.h(-1)) with ball dribbling. Thereafter, 11 players were submitted to 3 efforts of 30 minutes at 100, 105, and 110% of AT(Hoff). The AT(Hoff) corresponded to the speed relative to 3.5 mmol.L-1 lactate concentration. The speed relative to 4.0 mmol.L-1 was assumed to be AT(Hoff4.0), and the AT(HoffBI) was determined through bisegmented adjustment. For comparisons, Student's t-test, intraclass correlation coefficient (ICC), and Bland and Altman analyses were used. For reproducibility, ICC, typical error, and coefficient of variation were used. No significant difference was found between AT test and retest determined using different methods. A positive correlation was observed between AT(Hoff) and AT(Hoff4.0). The MLSSHoff (10.6 +/- 1.3 km.h(-1)) was significantly different compared with AT(Hoff) (10.2 +/- 1.2 km.h(-1)) and AT(HoffBI) (9.5 +/- 0.4 km.h(-1)) but did not show any difference from LAn(Hoff4.0) (10.7 +/- 1.4 km.h(-1)). The MLSSHoff presented high ICCs with AT(Hoff) and AT(Hoff4.0) (ICC = 0.94; and ICC = 0.89; p <= 0.05, respectively), without significant correlation with AT(HoffBI). The results suggest that AT determined on the Hoff circuit is reproducible and capable of predicting MLSS. The AT(Hoff4.0) was the method that presented a better approximation to MLSS. Therefore, it is possible to assess submaximal physiological variables through a specific circuit performed with the ball in young soccer players.

Pareto Frontier for the time-energy cost vector to an Earth-Moon transfer orbit using the patched-conic approximation

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

A radial basis function network (RBFN) for function approximation

A radial basis function network (RBFN) circuit for function approximation is presented. Simulation and experimental results show that the network has good approximation capabilities. The RBFN was a squared hyperbolic secant with three adjustable parameters amplitude, width and center. To test the network a sinusoidal and sine function,vas approximated.

An analog implementation of radial basis function network appropriate for transducer linearizer

A circuit for transducer linearizer tasks have been designed and built using discrete components and it implements by: a Radial Basis Function Network (RBFN) with three basis functions. The application in a linearized thermistor showed that the network has good approximation capabilities. The circuit advantages is the amplitude, width and center.

Comparative study between powers of sigmoid functions, MLP-backpropagation and polynomials in function approximation problems

Function approximation is a very important task in environments where the computation has to be based on extracting information from data samples in real world processes. So, the development of new mathematical model is a very important activity to guarantee the evolution of the function approximation area. In this sense, we will present the Polynomials Powers of Sigmoid (PPS) as a linear neural network. In this paper, we will introduce one series of practical results for the Polynomials Powers of Sigmoid, where we will show some advantages of the use of the powers of sigmiod functions in relationship the traditional MLP-Backpropagation and Polynomials in functions approximation problems.

Whaling error

You published recently (Nature 374, 587; 1995) a report headed "Error re-opens 'scientific' whaling debate". The error in question, however, relates to commercial whaling, not to scientific whaling. Although Norway cites science as a basis for the way in which it sets its own quota. scientific whaling means something quite different. namely killing whales for research purposes. Any member of the International Whaling Commission (IWC) has the right to conduct a research catch under the International Convention for the Regulation of Whaling. 1946. The IWC has reviewed new research or scientific whaling programs for Japan and Norway since the IWC moratorium on commercial whaling began in 1986. In every case, the IWC advised Japan and Norway to reconsider the lethal aspects of their research programs. Last year, however, Norway started a commercial hunt in combination with its scientific catch, despite the IWC moratorium.

Approximation Algorithms for Survivable Multicommodity Flow Problems with Applications to Network Design

Multicommodity flow (MF) problems have a wide variety of applications in areas such as VLSI circuit design, network design, etc., and are therefore very well studied. The fractional MF problems are polynomial time solvable while integer versions are NP-complete. However, exact algorithms to solve the fractional MF problems have high computational complexity. Therefore approximation algorithms to solve the fractional MF problems have been explored in the literature to reduce their computational complexity. Using these approximation algorithms and the randomized rounding technique, polynomial time approximation algorithms have been explored in the literature. In the design of high-speed networks, such as optical wavelength division multiplexing (WDM) networks, providing survivability carries great significance. Survivability is the ability of the network to recover from failures. It further increases the complexity of network design and presents network designers with more formidable challenges. In this work we formulate the survivable versions of the MF problems. We build approximation algorithms for the survivable multicommodity flow (SMF) problems based on the framework of the approximation algorithms for the MF problems presented in [1] and [2]. We discuss applications of the SMF problems to solve survivable routing in capacitated networks.

Applications of Linear Programming to Coding Theory

Maximum-likelihood decoding is often the optimal decoding rule one can use, but it is very costly to implement in a general setting. Much effort has therefore been dedicated to find efficient decoding algorithms that either achieve or approximate the error-correcting performance of the maximum-likelihood decoder. This dissertation examines two approaches to this problem. In 2003 Feldman and his collaborators defined the linear programming decoder, which operates by solving a linear programming relaxation of the maximum-likelihood decoding problem. As with many modern decoding algorithms, is possible for the linear programming decoder to output vectors that do not correspond to codewords; such vectors are known as pseudocodewords. In this work, we completely classify the set of linear programming pseudocodewords for the family of cycle codes. For the case of the binary symmetric channel, another approximation of maximum-likelihood decoding was introduced by Omura in 1972. This decoder employs an iterative algorithm whose behavior closely mimics that of the simplex algorithm. We generalize Omura's decoder to operate on any binary-input memoryless channel, thus obtaining a soft-decision decoding algorithm. Further, we prove that the probability of the generalized algorithm returning the maximum-likelihood codeword approaches 1 as the number of iterations goes to infinity.