Navegando por Assunto "Aprendizado do computador"

Agora exibindo 1 - 6 de 6

Acesso aberto (Open Access)
Análise de atributos de classificação para o diagnóstico de falhas em rolamentos baseado em SVM
(Universidade Federal do Pará, 2019-08-06) SOUZA, Jusley da Silva; BAYMA, Rafael Suzuki; http://lattes.cnpq.br/6240525080111166; MESQUITA, Alexandre Luiz Amarante; http://lattes.cnpq.br/3605920981600245; https://orcid.org/0000-0001-5605-8381
In industries, the concern in total availability of machines and the mechanical equipment in the productive area it’s subject of research and tests to obtain more efficient techniques to be applied for monitoring and faults’ diagnosing. Bearings are machine elements of great application in the industrial area and they present high fault index that generate machine’s stops to carry out maintenance. For this reason, this paper presents Artificial Intelligence technique applied to the vibration signals of a rotary machine for fault diagnosis in its bearings. The vibration signals are part of an open database offered by Case Western Reserve University. In this paper the Support Vector Machine (SVM) classification algorithm is applied in two ways for the rolling bearings faults’ diagnosis. In the first case statistical predictors (Root Mean Square Value, Crest Factor, K Factor, Kurtosis and Skewness) are used as features for the SVM classifier. In the second case, the signal processing is performed by applying the Ensemble Empirical Mode Decomposition (EEMD), which generates several signals called Intrinsic Mode Functions (IMFs). For each IMF, it’s modeled using Autoregressive Modeling (AR), and the AR modeling coefficients of each IMF are used as features for the SVM classifier. The analyzes are performed for training and validation groups, with randomly chosen window and with temporal sequence chosen window, considering two classification problems within the same data, the first one considers the same severity and only changes the fault type and the other vary both severity and fault type. As result, both methodologies presented excellent reliability results for bearing faults’ diagnosis.
Acesso aberto (Open Access)
Avaliação automática de questões discursivas usando LSA
(Universidade Federal do Pará, 2016-02-05) SANTOS, João Carlos Alves dos; FAVERO, Eloi Luiz; http://lattes.cnpq.br/1497269209026542
This work investigates the use of a model using Latent Semantic Analysis (LSA) In the automatic evaluation of short answers, with an average of 25 to 70 words, of questions Discursive With the emergence of virtual learning environments, research on Automatic correction have become more relevant as they allow the mechanical correction With low cost for open questions. In addition, automatic Feedback and eliminates manual correction work. This allows you to create classes With large numbers of students (hundreds or thousands). Evaluation research Texts have been developed since the 1960s, but only in the The current decade are achieving the necessary accuracy for practical use in teaching. For end users to have confidence, the research challenge is to develop Evaluation systems that are robust and close to human evaluators. despite Some studies point in this direction, there are still many points to be explored In the surveys. One point is the use of bigrasms with LSA, even if it does not contribute Very much with the accuracy, contributes with the robustness, that we can define as reliability2, Because it considers the order of words within the text. Seeking to perfect an LSA model In the direction of improving accuracy and increasing robustness we work in four directions: First, we include word bigrasms in the LSA model; Second, we combine models Co-occurrence of unigram and bigrams using multiple linear regression; third, We added a stage of adjustments on the LSA model score based on the Number of words of the responses evaluated; Fourth, we performed an analysis of the Of the scores attributed by the LSA model against human evaluators. To evaluate the We compared the accuracy of the system against the accuracy of human evaluators Verifying how close the system is to a human evaluator. We use a LSA model with five steps: 1) pre-processing, 2) weighting, 3) decomposition a Singular values, 4) classification and 5) model adjustments. For each stage it was explored Strategies that influenced the final accuracy. In the experiments we obtained An 84.94% accuracy in a comparative assessment against human Correlation among human specialists was 84.93%. In the field studied, the Evaluation technology had results close to those of the human evaluators Showing that it is reaching a degree of maturity to be used in Assessment in virtual learning environments. Google Tradutor para empresas:Google Toolkit de tradução para appsTradutor de sitesGlobal Market Finder.
Acesso aberto (Open Access)
Identificação de danos em estruturas usando modelo preditor baseado em técnicas de aprendizagem de máquinas
(Universidade Federal do Pará, 2019-10-04) BONA, Vanessa Cordeiro de; BAYMA, Rafael Suzuki; http://lattes.cnpq.br/6240525080111166; MESQUITA, Alexandre Luiz Amarante; http://lattes.cnpq.br/3605920981600245; https://orcid.org/0000-0001-5605-8381
The increase in the number of new buildings and the existence of countless old buildings, whether small or large, call attention to the need for measures that maintain the quality, safety and useful life of the structures. Inspections and monitoring, regardless of the age of the building, are essential to detect the existence of damage, especially in its initial phase, avoiding its propagation or serious consequences that originate due to a collapse of the structure, due to the high degree deterioration and no recovery techniques. Based on these aspects, this dissertation has the general objective of detecting damage in structures using the machine learning approach, which integrates three techniques: initially the Ensemble Empirical Mode Decomposition (EEMD) is applied a processing of the signals and seeks to adapt them for the application of the Auto Regressive Model (AR) generating the attributes, which will serve as input patterns for the Support Vector Machine (SVM) classifier. The data used to apply the methods come from the modeling of bi-supported steel beams, intact and with damaged regions, by the SAP 2000 Structural Analysis Software. With reference to the creation of the structures by finite elements, two types of loads were applied . The first case of random loading acting in only one point of the beam and the second case with three simultaneous loads in three points of the beam. According to variations in the location and degree of severity of the damage, the study sought to assess the ability of the predictive models to classify the data correctly. In the analyzes with greater mass losses, the accuracy values are higher, decreasing according to the reduction of the damage geometry, as the signs of displacement become similar to the integral structure. Regarding the number of loads, the method demonstrated better performance and accuracy in cases with three simultaneous loads.
Acesso aberto (Open Access)
Um modelo bayesiano que auxilia na identificação de alunos com dificuldades na aprendizagem de programação de computadores
(Universidade Federal do Pará, 2019-05-10) CAMPOS, Willys do Socorro Almeida de; TEIXEIRA, Otávio Noura; http://lattes.cnpq.br/5784356232477760; https://orcid.org/0000-0002-7860-5996; REIS, Rodrigo Quites; http://lattes.cnpq.br/9839778710074372; https://orcid.org/0000-0002-3657-4175
The learning of computer programming subjects has always brought challenges to any class of computer students, due to the difficulties linked to its use. This scenario greatly contributes to the demotivation of the student and, thus, the increased dropout of courses. Generally, teachers who work in these disciplines have signs about which students will become good programmers, but it is difficult to detect which need help in the learning process. This article proposes the use of a Bayesian model that helps in the identification of students with difficulties in the computer programming subjects. The research uses a mixed approach, quantitative and qualitative. An experiment, with an informal character, was carried out with students who were studying programming subjects. This set of students was presented to five specialist teachers in order to identify which ones would need help with the learning of programming. The same set was presented to the Bayesian model. The results showed that the model can help in the identification of students who present difficulties, with the potential to contribute to the learning process.
Acesso aberto (Open Access)
Modelo de rádio propagação em UHF para ambientes não homogêneos e climas distintos utilizando técnica de aprendizagem de máquina
(Universidade Federal do Pará, 2015-08-20) GOMES, Cristiane Ruiz; CAVALCANTE, Gervásio Protásio dos Santos; http://lattes.cnpq.br/2265948982068382
The digital TV broadcasts have greatly increased worldwide in recent years, especially in Brazil. The establishment and improvement of these transmission systems rely on models that take into account, among other factors, the geographical characteristics of the region, as these contribute to signal degradation. In Brazil, there is a great diversity of scenery and climates. For years several propagation models have been studied many for several frequency bands and types of paths. This thesis proposes an outdoor empirical radio propagation model for UHF band, which is used in digital TV. The proposed model estimates received power values can be applied to non-homogeneous paths and different climates, this latter presents innovative character for the UHF band. Different artificial intelligence techniques were chosen for theoretical and computational basis for having the ability to introduce, organize and describe quantitative and qualitative data quickly and efficiently, making it possible to determine the received power in a variety of settings and climates. The proposed model was applied to a city in the Amazon region with heterogeneous paths and wooded urban areas, fractions of freshwater among others. Measurement campaigns were conducted to obtain data signals from two digital TV stations in the metropolitan area of the city of Belém-Pará to model, compare and validate the model. The results are consistent. The model depicts a distinct difference between the two seasons of the studied year and small RMS errors for all cases studied. The validation of the model was based on the comparison with empirical and deterministic models.
Acesso aberto (Open Access)
Predição de comportamento de usuários oriundos do marketing digital por meio de redes neurais artificiais e aprendizado supervisionado
(Universidade Federal do Pará, 2019) ALVES, Vitor Pinheiro; TEIXEIRA, Otávio Noura; http://lattes.cnpq.br/5784356232477760; https://orcid.org/0000-0002-7860-5996
Success in attracting customers using marketing techniques creates a billionare problem wich is one of the most difficult that is selecting among the many prospects, which are more likely to become a customer. This work uses artificial neural networks to analyze the dataset generated from digital marketig techniques and classify which prospects have a greater chance to become a customer and which ones should be discarded. The Neural Network scores approximately 70% of cases among 3,541 records processed.