DSpace Coleção:
http://repositorio.ufc.br/handle/riufc/22000
2024-04-19T05:25:56ZImplantação de modelos de aprendizado de máquina no formato onnx utilizando diferentes frameworks
http://repositorio.ufc.br/handle/riufc/76653
Título: Implantação de modelos de aprendizado de máquina no formato onnx utilizando diferentes frameworks
Autor(es): Braga, Pedro Henrique Spinosa
Abstract: This paper presents a systematic analysis of the Open Neural Network Exchange (ONNX) and ONNXRuntime, along with the main frameworks for training and serving machine learning models. The general objective is to evaluate the performance and interoperability of ONNX Runtime on multiple platforms, both in batch and real-time inference, comparing it with other inference solutions. In addition, specific objectives are presented, including a literature review on ONNXandONNXRuntime, the development of a pipeline for distributed execution of tested algorithms, the evaluation of ONNX Runtime performance, the analysis of ONNX interoperability with different machine learning frameworks, and the documentation of the results of the systematic analysis and experiments. The study also includes a comparative analysis of fundamental characteristics among related works, such as inference engines, platforms used, application domains, evaluated metrics, models used, and hardware accelerators. The obtained results provide insights and conclusions on the performance and interoperability of ONNX Runtime, contributing to the understanding and improvement of deploying machine learning models in different environments and usage scenarios.
Tipo: TCC2023-01-01T00:00:00ZClassificação de arritmia cardíaca com aprendizado de máquina automatizado (AutoML)
http://repositorio.ufc.br/handle/riufc/76652
Título: Classificação de arritmia cardíaca com aprendizado de máquina automatizado (AutoML)
Autor(es): Miranda, Paula Luana Oliveira
Abstract: Cardiovascular diseases are the biggest killers in the world, with cardiac arrhythmia being one of the biggest causes. The test capable of detecting cardiac arrhythmia is the electrocardiogram, a test that contains noise and is difficult to interpret. In view of this, in order to contribute to research that facilitates the process of interpreting exams, Artificial Intelligence has a vast amount of work related to this objective, where researchers have developed strategies that make data more readable for algorithms and analyzed different model structures that better handle electrocardiogram data. Achieving the same objective, the present work implements and analyzes: segmentation sizes on Electrocardiogram (ECG) exam data; application of noise removal technique from the Neurokit library; Synthetic Oversampling Technique for Minority Classes (SMOTE) and Adaptive Synthetic Sampling Technique for Minority Classes (ADASYN); and uses automated machine learning to search for better hyperparameters that fit algorithms without manually establishing tests. Furthermore, the present work surveys the variations in hyperparameters generated for the models by the automated machine learning tool, AutoGluon. With the approaches implemented, this work concludes that for most of the algorithms used, the use of data with 360 segmentation allows the model to achieve greater performance when compared to data segmented in 2000. Machine learning models achieved better performances when classifying data that had the noise removal technique applied. Finally, Autogluon generated more variations for deep learning models, but despite this, the Light Gradient Boosting Machine (LightGBM) machine learning algorithm also achieved high performance in the testing phase. Therefore, the algorithms that achieved the best performance were neural networks and LightGBM.
Tipo: TCC2023-01-01T00:00:00ZO Uso de Aprendizado Profundo na classificação de ressonâncias magnéticas para detecção de tumor cerebral
http://repositorio.ufc.br/handle/riufc/76650
Título: O Uso de Aprendizado Profundo na classificação de ressonâncias magnéticas para detecção de tumor cerebral
Autor(es): Costa Junior, Francisco Valdemi Leal
Abstract: The proposed work aims to explore the application of computer vision techniques in the classification of MRI scans to facilitate the diagnosis of brain tumors. Early diagnosis of brain tumors is essential for appropriate treatment and better outcomes for patients. With the advancement of technology and the increasing use of medical images, the use of machine learning algorithms has shown promise in this field. The work is based on the use of two widely known architectures: Transformers and Convolutional (CNN). These architectures are capable of extracting relevant features from images, which is essential for detecting brain tumors using computer vision. The literature presents the models that obtained the best results from each architecture, along with image pre-processing techniques, which will be used to highlight relevant characteristics and reduce image noise. The aim is that this work contributes to advances in the early diagnosis of brain tumors through the application of computer vision techniques and shows the importance of using artificial intelligence in the health sector. The use of architectures such as Transformers and Convolutional, making a comparison between computer vision architectures.
Tipo: TCC2023-01-01T00:00:00ZLimites superiores para o número cromático L(3,2,1) de classes de grafos com grau máximo três
http://repositorio.ufc.br/handle/riufc/76649
Título: Limites superiores para o número cromático L(3,2,1) de classes de grafos com grau máximo três
Autor(es): Santos, Eliabe Soares
Abstract: Given a graph G, a function f :V(G) → {0,1,2,...,k} is said to be an L(3,2,1)-labelling of G if, for each u,v ∈V(G): when d(u,v) = 1, then |f(u)− f(v)| ≥ 3; when d(u,v) = 2, then | f (u) − f(v)| ≥ 2; and when d(u,v) = 3, then |f(u)− f(v)| ≥ 1. The span of an L(3,2,1)labelling f is the largest label assigned by f to a node of G. An L(3,2,1)-labelling is said to be optimal when it has the minimum possible span, and the span of such a labeling is called the L(3,2,1) chromatic number, and denoted by λ3,2,1(G). The class of cubic graphs is of interest in Graph Theory and Algorithms because many decision problems in arbitrary graphs are NPcomplete and remain NP-Complete even when constrained to cubic graphs. A graph is said to be subcubic when its maximum degree is equal to 3. Another class of graphs relevant to Graph Theory is the class of snark graphs. A snark graph is a bridgeless cubic graph that does not have a proper edge coloring with 3 colors. In this work, we show a tighter upper bound for the L(3,2,1) chromatic number of a subclass of subcubic graphs, such that there are no adjacent nodes with maximum degree such that any two nodes with degree equal to 3 are at distance at least 4 from each other, improving upon previous results in the literature. We also show tighter upper bounds for the L(3,2,1) chromatic number for the following infinite families of snarks: generalized Blanuša snarks, Loupekine snarks, and Flower snarks.
Tipo: TCC2023-01-01T00:00:00Z