FE - DI | Documentos por Auto-Depósito
Permanent URI for this collection
Browse
Recent Submissions
- Carbon Fiber Epoxy Composites for Both Strengthening and Health Monitoring of StructuresPublication . Salvado, Rita; Lopes, Catarina; Szojda, Leszek; Araújo, Pedro; Górski, Marcin; Velez, Fernando; Castro-Gomes, João; Krzywon, RafalThis paper presents a study of the electrical and mechanical behavior of several continuous carbon fibers epoxy composites for both strengthening and monitoring of structures. In these composites, the arrangement of fibers was deliberately diversified to test and understand the ability of the composites for self-sensing low strains. Composites with different arrangements of fibers and textile weaves, mainly unidirectional continuous carbon reinforced composites, were tested at the dynamometer. A two-probe method was considered to measure the relative electrical resistance of these composites during loading. The measured relative electrical resistance includes volume and contact electrical resistances. For all tested specimens, it increases with an increase in tensile strain, at low strain values. This is explained by the improved alignment of fibers and resulting reduction of the number of possible contacts between fibers during loading, increasing as a consequence the contact electrical resistance of the composite. Laboratory tests on strengthening of structural elements were also performed, making hand-made composites by the “wet process”, which is commonly used in civil engineering for the strengthening of all types of structures in-situ. Results show that the woven epoxy composite, used for strengthening of concrete elements is also able to sense low deformations, below 1%. Moreover, results clearly show that this textile sensor also improves the mechanical work of the strengthened structural elements, increasing their bearing capacity. Finally, the set of obtained results supports the concept of a textile fabric capable of both structural upgrade and self-monitoring of structures, especially large structures of difficult access and needing constant, sometimes very expensive, health monitoring.
- Leveraging Machine Learning for Weed Management and Crop Enhancement: Vineyard Flora ClassificationPublication . Corceiro, Ana; Pereira, Nuno José Matos; Alibabaei, Khadijeh; Gaspar, Pedro DinisThe global population’s rapid growth necessitates a 70% increase in agricultural production, posing challenges exacerbated by weed infestation and herbicide drawbacks. To address this, machine learning (ML) models, particularly convolutional neural networks (CNNs), are employed in precision agriculture (PA) for weed detection. This study focuses on testing CNN architectures for image classification tasks using the PyTorch framework, emphasizing hyperparameter optimization. Four groups of experiments were carried out: the first one trained all the PyTorch architectures, followed by the creation of a baseline, the evaluation of a new and extended dataset in the best models, and finally, the test phase was conducted using a web application developed for this purpose. Of 80 CNN sub-architectures tested, the MaxVit, ShuffleNet, and EfficientNet models stand out, achieving a maximum accuracy of 96.0%, 99.3%, and 99.3%, respectively, for the first test phase of PyTorch classification architectures. In addition, EfficientNet_B1 and EfficientNet_B5 stood out compared to all other models. During experiment 3, with a new dataset, both models achieved a high accuracy of 95.13% and 94.83%, respectively. Furthermore, in experiment 4, both EfficientNet_B1 and EfficientNet_B5 achieved a maximum accuracy of 96.15%, the highest one. ML models can help to automate crop problem detection, promote organic farming, optimize resource use, aid precision farming, reduce waste, boost efficiency, and contribute to a greener, sustainable agricultural future.
- Imputação de Valores Omissos em Análise Descritiva de Dados, em RPublication . Salambiaku, Luzizila; Prata, Paula; Ferrão, Maria EugéniaOs valores omissos representam um problema frequente no processo de análise de dados. Neste artigo foram comparados seis métodos distintos de imputação, disponíveis no software R e avaliado o seu desempenho em conjuntos de dados relacionados com a área da educação. Foi estudada uma amostra de 20408 estudantes para testar os seis algoritmos em quatro conjuntos de dados gerados por simulação com diferentes percentagens de valores omissos, considerando 5%, 10%, 15% e 20% nas variáveis de interesse. Foram explorados métodos de imputação simples (Média, Mediana e Moda), métodos baseados em aprendizagem automática (kNN e bPCA) e um método de imputação múltipla (MICE). Foi avaliado o desempenho de cada método calculando os respetivos erros de imputação através as métricas RMSE e MAE. Os resultados obtidos mostram que a imputação pela Moda forneceu quase de forma constante menores valores de erro.
- Anonymized Data Assessment via Analysis of Variance: An Application to Higher Education EvaluationPublication . Ferrão, Maria Eugénia; Prata, Paula; Fazendeiro, PauloThe assessment of the utility of an anonymized data set can be operational-ized by the determination of the amount of information loss. To investigate the possible degradation of the relationship between variables after anony-mization, hence measuring the loss, we perform an a posteriori analysis of variance. Several anonymized scenarios are compared with the original data. Differential privacy is applied as data anonymization process. We assess data utility based on the agreement between the original data structure and the anonymized structures. Data quality and utility are quantified by standard metrics, characteristics of the groups obtained. In addition, we use analysis of variance to show how estimates change. For illustration, we apply this ap-proach to Brazilian Higher Education data with focus on the main effects of interaction terms involving gender differentiation. The findings indicate that blindly using anonymized data for scientific purposes could potentially un-dermine the validity of the conclusions.
- Prediction of the Arbutus unedo colonization time via an agent-based distribution modelPublication . Bioco, João; Prata, Paula; Canovas, Fernando; Fazendeiro, PauloSpecies distribution models (SDMs) have been used to predict the distribution of species in an environment. Usually, this prediction is based on previous species occurrence data. Nowadays, several species are facing climatic changes that impact the way species behave. Global warming has implicated species displacement from their natural habitat to new suitable places. SDMs can be used to anticipate possible impacts of the climatic changes in species distribution, preventing species extinction scenarios. New SDMs approaches are needed to give valuable information that better approximates these models to reality. This paper presents a novel approach to the prediction of species distribution. It starts by using an agent-based model (ABM) to establish an approximate mapping between the geological and the computational times. Afterwards, the implemented ABM can be used to predict the species distribution at different time intervals. The presented case study concerns the distribution of the Arbutus unedo in the Iberian peninsula, a species with relevant socio-economic impact in Portugal. The results show that the measurement of the geological time, supported by the approximate correspondence to the number of epochs of an ABM, can be a valuable tool for the prediction of the species distribution in a changing environmental scenario.
- Teaching in conditions of difficult knowledge transfer due to the state of emergency caused by the pandemicPublication . Mravik, Miloš; Šarac, Marko; Veinovic, Mladen; Pombo, NunoIntroduction/purpose: This paper presents the transformation of the current, classical approach to teaching. Online platforms enable students with and without disabilities to follow classes without hindrance during the lecture period. After the lecture, they are allowed to view video and presentation materials. The main advantage of this way of teaching is the possibility of attending classes from any location and from any device; it is only important to be connected to the Internet. Methods: Full integration with the already existing Faculty Information System has been performed. The paper describes a new approach to teaching and illustrates the expected benefits of online teaching. The platforms used in this integration are Microsoft Azure, Microsoft Office 365 Admin, Microsoft Teams, Microsoft Stream and Microsoft SharePoint. Results: The result of the test of work with students showed that by introducing a system for online teaching, we directly affect the improvement and quality of teaching. Conclusion: Considering all the results, it can be concluded that the transition to the online way of teaching allows end listeners a comprehensive transfer of knowledge as well as re-listening to the same. This model can be used for an unlimited number of users in all Institutions, regardless of whether the field of activity of these Institutions is of educational origin.
- Utility-driven assessment of anonymized data via clusteringPublication . Ferrão, Maria Eugénia; Prata, Paula; Fazendeiro, PauloIn this study, clustering is conceived as an auxiliary tool to identify groups of special interest. This approach was applied to a real dataset concerning an entire Portuguese cohort of higher education Law students. Several anonymized clustering scenarios were compared against the original cluster solution. The clustering techniques were explored as data utility models in the context of data anonymization, using k-anonymity and (ε, δ)-differential as privacy models. The purpose was to assess anonymized data utility by standard metrics, by the characteristics of the groups obtained, and the relative risk (a relevant metric in social sciences research). For a matter of self-containment, we present an overview of anonymization and clustering methods. We used a partitional clustering algorithm and analyzed several clustering validity indices to understand to what extent the data structure is preserved, or not, after data anonymization. The results suggest that for low dimensionality/cardinality datasets the anonymization procedure easily jeopardizes the clustering endeavor. In addition, there is evidence that relevant field-of-study estimates obtained from anonymized data are biased.
- On the Modelling of Species Distribution: Logistic Regression Versus Density Probability FunctionPublication . Bioco, João; Prata, Paula; Canovas, Fernando; Fazendeiro, PauloThe concerns related to climate changes have been gaining attention in the last few years due to the negative impacts on the environment, economy, and society. To better understand and anticipate the effects of climate changes in the distribution of species, several techniques have been adopted comprising models of different complexity. In general, these models apply algorithms and statistical methods capable of predicting in a particular study area, the locations considered suitable for a species to survive and reproduce, given a set of eco-geographical variables that influence species behavior. Logistic regression algorithm and Probability density function are two common methods that can be used to model the species suitability. The former is a representative of a class of models that requires the availability (or imputation) of presence-absence data whereas the latter represents the models that only require presence data. Both approaches are compared regarding the capability to accurately predict the environmental suitability for species. On a different way, the behaviour of the species in the projected environments are analysed by simulating its potential distribution in the projected environment. A case study reporting results from two types of species with economical interest is presented: the strawberry tree (Arbutus unedo) in mainland Portugal, and the Apis mellifera (African Lineage) in the Iberian Peninsula.
- SDSim: A generalized user friendly web ABM system to simulate spatiotemporal distribution of species under environmental scenariosPublication . Bioco, João; Cánovas, Fernando; Prata, Paula; Fazendeiro, PauloThis paper presents the Agent-Based Modelling System of spatial distribution of species SDSim. SDSim is an agent-based modelling system designed to simulate spatial distribution of species and populations for conservation and management purposes. SDSim gives to modellers the ability to simulate movements and colonization patterns of species given locations under study and a set of eco-geographical variables in which species depends on.
- Synchronization Overlap Trade-Off for a Model of Spatial Distribution of SpeciesPublication . Bioco, João; Prata, Paula; Cánovas, Fernando; Fazendeiro, PauloDespite of the widespread implementation of agent-based models in ecological modeling and another several areas, modelers have been concerned by the time consuming of these type of models. This paper presents a strategy to parallelize an agent-based model of spatial distribution of biological species, operating in a multi-stage synchronous distributed memory mode, as a way to obtain gains in the performance while reducing the need for synchronization. A multiprocessing implementation divides the environment (a rectangular grid corresponding to the study area) into stage-subsets, according to the number of defined or available processes. In order to ensure that there is no information loss, each stage-subset is extended with an overlapping section from each one of its neighbouring stage-subsets. The effect of the size of this overlapping on the quality of the simulations is studied. These results seem to indicate that it is possible to establish an optimal trade-off between the level of redundancy and the synchronization frequency. The reported paralellization method was tested in a standalone multicore machine but may be seamlessly scalable to a computation cluster.