TCC - Trabalho de Conclusão de Curso http://hdl.handle.net/123456789/11 2026-04-17T16:02:16Z 2026-04-17T16:02:16Z SILVA, RAFAEL RIBEIRO DA http://hdl.handle.net/123456789/378 2019-06-12T19:24:58Z 2018-11-19T00:00:00Z

DEEP LEARNING APLICADO À PREDIÇÃO DE TENDÊNCIAS NO MERCADO DE AÇÕES SILVA, RAFAEL RIBEIRO DA Aiming to achieve the goal of training neural networks to enable them to identify trends in the price of a stock, data were collected on the stocks of 7,145 companies during the period of one month. These data were transformed into candlestick charts that, using technical analysis concepts, were classified as being charts with high, low and consolidation tendencies and formed the bases used to train, validate and test ResNet and Xception network models. It was possible to conclude that both obtained satisfactory results, but with advantage to ResNet, that obtained greater precisions from the training phase until the final tests. In the final tests ResNet reached 87% of accuracy and F1 Score, while Xception reached 86%.

2018-11-19T00:00:00Z SILVA, WILLIAM SDAYLE MARINS http://hdl.handle.net/123456789/373 2019-06-12T18:56:16Z 2018-11-26T00:00:00Z

DETECÇÃO DE FOGO EM FRAMES DE VÍDEOS UTILIZANDO APRENDIZAGEM DE MÁQUINA SILVA, WILLIAM SDAYLE MARINS Fires can occur at any time, just having the objects needed for it, so they can occur anywhere leaving several problems wherever it goes. Based on this idea, this work was proposed to perform a fire detection in videos, using convolutional neural networks to validate the image of the videos. A base with videos was created, containing videos with similar characteristics, those of fire and smoke, from which were generated images containing only the movement that was detected in the video. After the extraction of the backgrounds, the images only contained parts of interest, after that, the images were used in convolutional neural networks, which the architectures were the Xception and ResNet. The results presented by both networks were extremely satisfactory for the propose of this work.

2018-11-26T00:00:00Z SILVA http://hdl.handle.net/123456789/372 2019-06-12T18:47:46Z 2018-11-19T00:00:00Z

RAFAEL RIBEIRO DA SILVA Aiming to achieve the goal of training neural networks to enable them to identify trends in the price of a stock, data were collected on the stocks of 7,145 companies during the period of one month. These data were transformed into candlestick charts that, using technical analysis concepts, were classified as being charts with high, low and consolidation tendencies and formed the bases used to train, validate and test ResNet and Xception network models. It was possible to conclude that both obtained satisfactory results, but with advantage to ResNet, that obtained greater precisions from the training phase until the final tests. In the final tests ResNet reached 87% of accuracy and F1 Score, while Xception reached 86%.

2018-11-19T00:00:00Z PINHO JÚNIOR, EVANDRO LUÍZ DE http://hdl.handle.net/123456789/371 2019-06-12T18:44:49Z 2018-11-21T00:00:00Z

AUTOMATIZAÇÃO DO SERVIÇO DE ATENDIMENTO AO CANDIDATO DO VESTIBULAR PINHO JÚNIOR, EVANDRO LUÍZ DE Chatbots can be used to automate repetitive activities related to customer service. In UENP entrance exam website, several questions are sent to the responsible e-mail for the service, several of them similar in purpose. The objective of this project is build a chatbot to the exam site, with the purpose of assist the current attendance system (by e-mail). In the development stage, several alternatives available in the market for the creation of a ranking with the most appropriate technologies were analyzed. The chatbot tests were carried out with several teams with different knowledge about the college entrance examination, and based on the results the conclusions about the project and the analyzed platforms were drawn.

2018-11-21T00:00:00Z