Armando Vieira

Hazy.com

Discover the potential applications, challenges, and opportunities of deep learning from a business perspective with technical examples. These applications include image recognition, segmentation and annotation, video processing and annotation, voice recognition, intelligent personal assistants, automated translation, and autonomous vehicles. An...

Recent advances in machine learning allow the design of algorithms that can learn directly from data without being manually designed. These cognitive algorithms are getting better and better, which can be seen at international challenges, as they are showing performance and accuracy superior to humans. On the other hand, as the amount of medical da...

Increases in throughput and installed base of biomedical research equipment led to a massive accumulation of -omics data known to be highly variable, high-dimensional, and sourced from multiple often incompatible data platforms. While this data may be useful for biomarker identification and drug discovery, the bulk of it remains underutilized. Deep...

We propose a robust classifier to predict buying intentions based on user behaviour within a large e-commerce website. In this work we compare traditional machine learning techniques with the most advanced deep learning approaches. We show that both Deep Belief Networks and Stacked Denoising auto-Encoders achieved a substantial improvement by extra...

this dataset was used for several publications on bankruptcy prediction. It contains 30 financial ratios of hundreds of french comanies and a state of the company (0 = ok, 1 = bankrupt) in the following year (2007)

We report on the effect of Pt on the growth kinetics of δ-Ni2Si and Ni1−xPtxSi thin films formed by solid phase reaction of a Ni(Pt) alloyed thin film on Si(100). The study was performed by real-time Rutherford backscattering spectrometry examining the silicide growth rates for initial Pt concentrations of 0, 1, 3, 7, and 10 at. % relative to the N...

Bankruptcy trajectory reflects the dynamic changes of financial situation of companies, and hence make possible to keep track of the evolution of companies and recognize the important trajectory patterns. This study aims at a compact visualization of the complex temporal behaviors in financial statements. We use self-organizing map (SOM) to analyze...

We report on the effect of Pt on the growth kinetics of d-Ni2Si and Ni1�xPtxSi thin films formed by solid phase reaction of a Ni(Pt) alloyed thin film on Si(100). The study was performed by real-time Rutherford backscattering spectrometry examining the silicide growth rates for initial Pt concentrations of 0, 1, 3, 7, and 10 at. % relative to the N...

Default risk models have lately raised a great interest due to the recent world economic crisis. In spite of many advanced techniques that have extensively been proposed, no comprehensive method incorporating a holistic perspective has hitherto been considered. Thus, the existing models for bankruptcy prediction lack the whole coverage of contextua...

We propose a credit scoring algorithm based on the supervised ISOMAP to rate SME. By projecting the companies balance sheet data into a one dimensional component we obtain a smoother distribution of ratings while increasing the discriminatory capability of each rate in terms of the probability of default. The method is applied to a large dataset of...

In this work a Multi-Objective Evolutionary Algorithm (MOEA) was applied for feature selection in the problem of bankruptcy prediction. This algorithm maximizes the accuracy of the classifier while keeping the number of features low. A two-objective problem, that is minimization of the number of features and accuracy maximization, was fully analyze...

We report on the solid-phase reaction of thin Ni-rare earth films on a Si(100) substrate, for a variety of rare earth (RE) elements (Y, Gd, Dy, and Er). Both interlayer (Ni/RE/〈Si〉) and alloy (Ni-RE/〈Si〉) configurations were studied. The phase sequence during reaction was revealed using real-time x-ray diffraction whereas the elemental diffusion an...

In this work a Multi-Objective Evolutionary Algorithm (MOEA) was applied for feature selection in the problem of bankruptcy prediction. This algorithm maximizes the accuracy of the classifier while keeping the number of features low. A two-objective problem, that is minimization of the number of features and accuracy maximization, was fully analyze...

As one of the major business problems, corporate bankruptcy has been extensively studied using a large variety of statistical and machine learning approaches. However, the trajectory of bankruptcy behavior is seldom explored in the literature. In this paper, we use self-organizing map neural networks to analyze the changes of financial situation of...

The prediction of bankruptcy is of significant importance with the present-day increase of bankrupt companies. In the practical applications, the cost of misclassification is worthy of consideration in the modeling in order to make accurate and desirable decisions. An effective prediction system requires the integration of the cost preference into...

Credit rating is involved in many financial applications to estimate the creditworthiness of corporations or individuals. In addition to building accurate credit rating models, the stability of models is of significant importance to economic performance. In this work we propose a methodology based on learning vector quantization (LVQ) to develop a...

Cost-sensitive learning is an important topic in bankruptcy prediction concerning the unequal misclassification cost of different classes. Learning vector quantization (LVQ) is a powerful tool to solve bankruptcy prediction problem as a classification task. The heuristic algorithms are applied widely in conjunction with artificial intelligent metho...

A Multi-Objective Evolutionary Algorithm (MOEA) was adapted in order to deal with problems of feature selection in data-mining. The aim is to maximize the accuracy of the classifier and/or to minimize the errors produced while minimizing the number of features necessary. A Support Vector Machines (SVM) classifier was adopted. Simultaneously, the pa...

The importance of cost-sensitive learning becomes crucial when the costs of misclassifications are quite different. Many evidences have demonstrated that a cost-sensitive predictive model is more desirable in practical applications than a traditional one without taking the cost into consideration. In this paper, we propose two approaches which inco...

In this work a Multi-Objective Evolutionary Algorithm (MOEA) was applied for feature selection in the problem of bankruptcy prediction. The aim is to maximize the accuracy of the classifier while keeping the number of features low. A two-objective problem - minimization of the number of features and accuracy maximization – is fully analyzed using t...

Financial distress prediction is of great importance to all stakeholders in order to enable better decision-making in evaluating firms. In recent years, the rate of bankruptcy has risen and it is becoming harder to estimate as companies become more complex and the asymmetric information between banks and firms increases. Although a great variety of...

This paper reports on the advantage of using artificial neural networks (ANNs) to analyze large sets of real-time Rutherford backscattering spectrometry (RBS) data. Real-time RBS, i.e. collecting RBS spectra at periodic time intervals during a thermal treatment, probes the full response of a thin film to the annealing in situ. Although very valuabl...

Cost-sensitive classification algorithms that enable effective prediction, where the costs of misclassification can be very different, are crucial to creditors and auditors in credit risk analysis. Learning vector quantization (LVQ) is a powerful tool to solve bankruptcy prediction problem as a classification task. The genetic algorithm (GA) is app...

Financial distress prediction is of crucial importance in credit risk analysis with the increasing competition and complexity of credit industry. Although a variety of methods have been applied in this field, there are still some problems remained. The accurate and sensitive prediction in presence of unequal misclassification costs is an important...

Prediction of financial distress of companies is analyzed with several machine learning approaches. We used Diane, a large database containing financial records from small and medium size French companies, from the year of 2002 up to 2007. It is shown that inclusion of historical data, up to 3 years priori to the analysis, increases the prediction...

Bankruptcy prediction is of great importance in financial statement analysis to minimize the risk of decision strategies. It attempts to separate distress companies from healthy ones according to some financial indicators. Since the real data usually contains irrelevant, redundant and correlated variables, it is necessary to reduce the dimensionali...

We report on a real-time Rutherford backscattering spectrometry study of the erratic redistribution of Pt during Ni silicide formation in a solid phase reaction. The inhomogeneous Pt redistribution in Ni(Pt)Si films is a consequence of the low solubility of Pt in Ni ₂ Si compared to NiSi and the limited mobility of Pt in NiSi. Pt further...

Cost-sensitive learning is of critical importance in many domains including bankruptcy prediction where the costs of different errors are unequal. Most existing classification methods aim to minimize overall error based on the assumption that the costs are equal. This paper presents three cost-sensitive learning vector quantization (LVQ) approaches...

In this paper we study personal credit scoring using several machine learning algorithms: Multilayer Perceptron, Logistic Regression, Support Vector Machines, AddaboostM1 and Hidden Layer Learning Vector Quantization. The scoring models were tested on a large dataset from a Portuguese bank. Results are benchmarked against traditional methods under...

We apply manifold learning to a real data set of distressed and healthy companies for proper geometric tunning of similarity data points and visualization. While Isomap algorithm is often used in unsupervised learning our approach combines this algorithm with information of class labels for bankruptcy prediction. We compare prediction results with...

In this paper we propose a supervised version of the Isomap algorithm by incorporating class label information into a dissimilarity matrix in a financial analysis setting. On the credible assumption that corporates financial status lie on a low dimensional manifold, nonlinear dimensionality reduction based on manifold learning techniques has strong...

Comparison of ternary fission of the metallic cluster Na27++ + into equal fragments with binary fission is made by use of the Shell Correction Method. We conclude that favourable fission barriers and energy releases of that tripartition along with some dynamical properties make ternary fission a very attractive process for further investigatio...

The Physics Teacher 44(7), 406 (2006) DOI: http://doi.org/10.1119/1.2353570

Making inferences and choosing appropriate responses based on incomplete, uncertainty and noisy data is challenging in financial settings particularly in bankruptcy detection. In an increasingly globalized economy, bankruptcy results both in huge economic losses and tremendous social impact. While early prediction for a bankruptcy, if done appropri...

Sol–gel processing is a cheap and versatile method of producing silica on silicon films for planar integrated optics. It leads to high Er incorporation and to easy incorporation of Ag, that can intensify the rare-earth photoluminescence. Different heat treatments and compositions must be tested to optimise the properties of the films grown, and the...

We have developed artificial neural networks (ANNs) for simultaneous analysis of Rutherford backscattering spectrometry and elastic recoil detection analysis data. The ANNs developed were applied to a highly complex problem, namely the analysis of multilayered silica-titania films doped with Ag and Er, where eleven parameters are required to descri...

Most problems in physics textbooks are highly idealized to keep them analytically manageable. However, in dealing with daily phenomena, some models presented in textbooks are oversimplified. The discrepancy between what students observe and what these models predict may cause frustration or even distrust. On the other hand, it is crucial to develop...

In this paper we apply several learning machine techniques to the problem of financial distress classification of medium-sized private companies. Financial data was obtained from Diana, a large database containing financial statements of French companies. Classification accuracy is evaluated with Artificial Neural Networks, Classification and Regre...

In this work we calculate the tension and equivalent resistance on a electric line feeding N equidistant loads resistances with continuous current. We obtain simple expressions using the approximation of a continuous line and considering the cable resistance small compared with the load resistance.

In this work we calculate the tension and equivalent resistance on a electric line feeding N equidistant loads resistances with continuous current. We obtain simple expressions using the approximation of a continuous line and considering the cable resistance small compared with the load resistance.

We present an Iterative Artificial Neural Network (IANN) in which computation is performed through a set of successive layers sharing the same weights. This network requires fewer weights while it can handle high-dimensional inputs. IANN is applied, with good results, to a time series prediction and two classification problems.

We used Artificial Neural Network for protein loop classification based on the amino acid sequence alone. A new algorithm recently proposed, the Hidden Layer Learning Vector Quantization (HLVQ) was used and its accuracy compared with traditional Multilayer Preceptrons (MLP). The HLVQ algorithm achieved superior accuracy correctly classifying most l...

We have previously developed artificial neural networks (ANN) dedicated to the analysis of RBS spectra. One of the limitations of the ANNs so far developed was that one single spectrum could be analysed from each sample. When more than one spectrum is collected, each had to be analysed separately, leading to different results and hence reduced accu...

Two different methods to accelerate the search of a Multi-Objective Evolutionary Algorithm (MOEA) using Artificial Neural Networks are presented. Two different methods are proposed. One using ANN to approximate the fitness of the solutions alternated with the real fitness evaluation, being the ANN approximation used only when the estimated error of...

The Hidden Layer Learning Vector Quantization is used to correct the prediction of multilayer perceptrons in classification of high-dimensional data. Corrections are significant for problems with insufficient training data to constrain learning. Our method, HLVQ-C, allows the inclusion of a large number of attributes without compromising the genera...

We report a generalisation of previous works where artificial neural networks (ANNs) were successfully applied for specific implantations such as Er in sapphire or Ge in Si. We have now developed a code that it is able to analyse data from implantations of any element with Z between 18 and 83 into any target composed of one or two lighter elements....

Predicting the financial health of companies is a problem of great importance to various stakeholders in the increasingly globalized economy. We apply several learning machines methods to the problem of bankruptcy prediction of private companies. Financial data obtained from Diana, a database containing 780,000 financial statements of French compan...

We present two methods to accelerate the search of a Multi-Objective Evolutionary Algorithm (MOEA) using Arti£- cial Neural Networks (ANN) to approximate the £tness functions. This approach can substantially reduce the number of £tness evalu- ations on computational expensive problems without compromising the good search capabilities of MOEA. In on...

In multilayered Ti0.4Al0.6N/Mo coatings, a strengthening effect can be obtained by using alternate layers of materials with high and low elastic constants. This behaviour requires a multilayer periodicity below a certain value in order to reduce dislocation motion across layer interface. Below this critical period, in most cases the hardness decrea...

Predicting the failure of a company is a difficult problem traditionally performed by accounting experts using heuristic rules extracted from experience. In this work we apply HLVQ, a new algorithm to train neural networks, to this problem and compared its results with G-Prop, a neural network optimized with evolutionary algorithms. We show that HL...

AlO(x)N(y) ultrathin films are used as insulating layers in advanced microelectronic devices. Structural characterization of these films is often done by the Rutherford backscattering (RBS) analysis. The RBS analysis of these oxinitrides is a difficult task since the relevant signals of the spectrum are washed out by the large substrate background...

We propose an algorithm for training multi layer preceptrons (MLP) for classi"cation prob- lems, that we named hidden layer learning vector quantization. It consists of applying learning vector quantization to the last hidden layer of a MLP and it gave very successful results on problems containing a large number of correlated inputs. It was applie...

We present an algorithm based on artificial neural networks able to determine optimized experimental conditions for Rutherford backscattering measurements of Ge-implanted Si. The algorithm can be implemented for any other element implanted into a lighter substrate. It is foreseeable that the method developed in this work can be applied to still man...

A tropical strain of Cryptomonas obovata Skuja, isolated from a shallow oxbow lake,releaseda sulfated fucose-rich polysaccharide. The polysaccharide is composed mainly offucose (42%), N-acetyl-galactosamine (26%) and rhamnose (15%), with smallquantities of glucuronic acid, mannose, galactose, xylose and glucose. Sulfateaccounted for 1.7% total poly...

Five elements (Ti, Fe, Co, Er and Au) were implanted in sapphire to fluences between 8×1013 and 5×1017 at/cm2, and energies between 200 and 800 keV. We used Rutherford backscattering to determine the dose and depth of the implanted elements. The data analysis is performed using an artificial neural network (ANN). Here we report a generalisation of...

We present an algorithm based on artificial neural networks (ANNs) able to determine optimised experimental conditions for Rutherford backscattering (RBS) measurements of Ge-implanted Si. The algorithm can be implemented for any element implanted into a lighter substrate, and can be extended to other ion beam analysis techniques. It is a push-butto...

A neural network algorithm has been successfully used to analyse proton elastic backscattering (EBS) data. The composition of NiTaC films deposited on Si substrates was determined. We show that, after being trained, the neural network can analyse the spectra from these samples with an excellent accuracy. Although there are other methods to analyse...

Rutherford backscattering spectrometry (RBS) is a well-established technique for the elemental depth profile of the surface layers of samples, including the determination of the dose and depth of implanted elements. We have developed a code based on artificial neural networks (ANN) to analyse RBS data. The ANN was trained using the traditional back...

The work presented in this paper is related with the development and integration of important communication, and creation, tools that allowed the authors to build a flexible and modular system dedicated to distance learning by computer communication. The distance learning system supports many facilities for publication, storage and dissemination of...

We have developed a code based on artificial neural networks (ANN) to analyse Rutherford backscattering data. In particular, we have applied the code to the analysis of germanium implants in silicon substrates. Here, we study the reliability and accuracy of the quantitative results obtained. We first constructed three different training data sets....

We used the Monte Carlo program MCNP to calculate the neutron and gamma fluxes on a fast neutron irradiation facility being installed on the Portuguese Research Reactor (RPI). The purpose of this facility is to provide a fast neutron beam for irradiation of electronic circuits. The gamma dose should be minimized. This is achieved by placing a lead...

A study of several commercial instrumentation amplifiers (INA110, INA111, INA114, INA116, INA118 & INA121) under neutron and vestigial gamma radiation was done. Some parameters (Gain, input offset voltage, input bias currents) were measured on-line and bandwidth, and slew rate were determined before and after radiation. The results of the testing o...

We developed a m ethod for fast optimisation of energy spectrum of neutron beams for BNCT using Artificial Neural Networks (ANN). The ANN was trained using a database containing 1000 simulations of the received RBE-doses on a Snyder phantom by the tumour and healthy tissues, for various neutron beams. The calculations were performed with the Monte...

The Monte Carlo program MCNP-4B was used to simulate the vertical access of the thermal column of the Portuguese Research Reactor. The reactor core was included in the simulations and two models for the neutron source were studied. Measurements were done with activation detectors. The order of magnitude of the thermal flux and gamma dose at the ver...

A neural network algorithm dedicated to recognition of Rutherford backscattering (RBS) data was developed. The algorithm was applied to one important particular case, namely the determination of the amount of Ge implanted in Si samples and the depth at which the Ge is located. An average error on both Ge amount and depth of less than 3% could be re...

Instrumentation amplifiers and voltage controlled current sources have been designed by using single operational amplifiers (OPA124, TLE2071 and OPA627). Their performance has been evaluated in the laboratory and under neutron radiation. On line measurements of the offset voltages, offset currents, closed loop gain, CMRR and bias currents were perf...

We use the liquid drop model, with the stabilized jellium model, and the Strutinsky shell correction method [Nucl. Phys. A 95, 420 (1967); 122, 1 (1968)], with the two-center asymmetric deformed harmonic-oscillator potential, to evaluate fission barriers for three representative simple metal clusters (sodium, aluminum, and potassium). We obtain fis...

We compare Kohn-Sham results (density, cohesive energy, size and effect of charging) of the Spherical Averaged Pseudopotential Model with the Stabilized Jellium Model for clusters of sodium and aluminum with less than 20 atoms. We find that the Stabilized Jellium Model, although conceptually and practically more simple, gives better results for the...

We used the Shell Correction Method to evaluate systematically fission barriers for charged metallic clusters of sodium and aluminum. Shell effects are responsible for the deviation of the fission barrier height from the value predicted by the Liquid Drop Model. Fragment deformations are essential degrees of freedom. The sizes for which fission co...