Detección de técnicas de aprendizaje profundo aplicadas en las diferentes áreas del conocimiento, empleando el método de revisión sistemática de literatura.

Deep Learning is a field of research that has become very popular in recent years for data learning, as it reaches high-level abstractions through the modeling of several layers of processing. Deep Learning has inspired a large number of researchers who use deep algorithms to detect patterns and extract data characteristics, in order to obtain more accurate results than commonly used models. That is why, thanks to the growing application of Deep Learning, the present Degree Work aims to determine the techniques and / or deep learning models that are currently applied in the areas of knowledge of engineering, medicine, research, industry and finance. For this, a Systematic Literature Review was developed based on the protocol of Barbara Kitchenham, resulting in the selection of 64 primary studies from which the research problem and the applied Deep Learning model were extracted. In the Systematic Literature Review it was obtained that, the Deep Learning LSTM model is ideal for problems of sales prediction, to validate this result the sales data of ILE (Lojana's Industry of Especerías) were used in developing a prediction model of sales using the LSTM network, which was compared with the models of Machine Learning simple neural network and the ARIMA model, using as an evaluation metric the mean square error, the experiments showed that despite the small size of the data the LSTM model gets less error than the other models. Finally, the DL has a wide range of applications, among them the sales prediction, where it was found that the LSTM model has a lower error than the Machine Learning models, although the data is limited. Therefore, the LSTM model is more efficient than Machine Learning models, although to validate the effectiveness of DL in future work, it is recommended to compare the results with other DL models.

• 主要作者: Suntaxi Sarango, Martha Cristina (author)
• 格式: bachelorThesis
• 语言: spa
• 出版: 2019
• 主题: <INTELIGENCIA ARTIFICIAL><APRENDIZAJE PROFUNDO><REDES NEURONALES ARTIFICIALES><DEEP LEARNING><RED LSTM>
• 在线阅读: http://dspace.unl.edu.ec/jspui/handle/123456789/21918