Repositório UFSCar

For more information, visit: RI UFSCar

Instructions for authors: Click here

Explore by document type

Discover UFSCar’s academic and scientific output

Communities

Select a community to browse its collections.

View all

Recent Submissions

  • Thumbnail Image
    Item type:Item,
    Métodos alternativos de inferência para modelos psicométricos de variáveis latentes de alta dimensão
    (Universidade Federal de São Carlos, 2026-09-04)
    Barros, Flavio Margarito Martins de
    ;
    Cúri, Mariana
    ;
    http://lattes.cnpq.br/7025653928429545
    ;
    https://orcid.org/0000-0002-7651-1064
    ;
    http://lattes.cnpq.br/1316455783756401
    Cognitive diagnostic models (CDMs), such as DINA and DINO, classify respondents according to multiple discrete latent attributes, providing interpretable diagnostic feedback. This interpretability comes at a computational cost: classical estimation by marginal maximum likelihood with the EM algorithm requires a sum over the 2^K latent profiles, which grows exponentially with the number of attributes K and makes inference impractical precisely in the high-dimensional settings where diagnosis would be most valuable. This thesis develops and evaluates scalable variational inference methods, based on automatic-differentiation optimisation, for the DINA and DINO models in high dimension, while preserving the psychometric structure of these models: the Q matrix, the ideal response function, the interpretable guessing gj and slipping sj item parameters, and the item-level monotonicity constraint gj + sj < 1. Two methodological families are proposed. The first is a relaxed, amortised variational autoencoder (VAE), in which a shared inference network maps the response vector to posterior mastery probabilities, and a structured decoder implements the DINA or DINO response function in differentiable form. The second is a local, non-amortised black-box variational inference (BBVI), with per-respondent variational parameters, considered in two variants: a score-function (REINFORCE) variant, which preserves binary samples, and a continuous-relaxation (Binary-Concrete) variant. The differentiable decoders are built from the product t-norm, for the conjunctive DINA rule, and the probabilistic t-conorm, for the disjunctive DINO rule. Replacing the categorical posterior over 2^K profiles with K Bernoulli factors per respondent, together with stochastic gradient estimation, avoids explicit enumeration and enables estimation in dimensions where enumeration is impractical. The methods are evaluated through simulation studies that vary the numbers of respondents, items and attributes, including high-dimensional and stress scenarios, assessing the recovery of item parameters and latent attributes, the variability across replications, and the computational time; a practical comparison with classical estimators implemented in R packages, at feasible dimensions, is also included. The results show that item-parameter recovery is strong in the evaluated scenarios with more items per attribute and that marginal attribute accuracy remains high, although full-profile recovery decays with K. Among the BBVI variants, the relaxed one shows shorter runtime and lower dispersion across replications in the evaluated configurations, whereas the score-function one avoids the continuous-relaxation bias at the cost of higher variance; the VAE, in turn, amortises inference for new respondents. The two families prove to be complementary: the work preserves the interpretability of CDMs while enabling scalable inference through variational approximations, automatic differentiation and accelerated computation.
  • Thumbnail Image
    Item type:Item,
    Da universidade para as escolas: produção de material para divulgação científica sobre o microplástico
    (Universidade Federal de São Carlos, 2025-12-17)
    Souza, Caroline Martins
    ;
    Anieli Cristina Maraschi
    ;
    http://lattes.cnpq.br/5896110724916724
    ;
    https://orcid.org/0000-0001-8653-9709
    ;
    Pierson, Alice Helena Campos
    Plastic pollution, especially from microplastics, has become one of the most significant environmental challenges today and highlights the search for strategies that bring science closer to society. However, most scientific dissemination materials prioritize results, and little is found about how scientific knowledge is produced, which hampers critical understanding and favors the spread of misinformation. In this context, the present study aims to develop scientific dissemination materials on microplastics that not only present the environmental impacts of this pollutant but also integrate elements of the scientific production process. For this purpose, bibliographic research was carried out to identify materials that articulated scientific dissemination, knowledge production, and the theme of plastics. Although several materials presented updated information about microplastics, no materials were found that described how studies are conducted within universities, revealing an essential gap in scientific communication. Based on this, four scientific dissemination materials were produced: two folders and two videos, all with accessible language, visual resources, and QR codes, to promote integration of the content. In the materials, the aim was to bring the reader closer to scientific work in laboratories while also explaining how research is conceived, funded, conducted, and translated into articles, reports, and publications. In addition, topics such as the origin and types of plastics, the formation of microplastics and nanoplastics, and the environmental and social impacts of plastics were addressed. Thus, the materials produced aim to foster scientific literacy, stimulate critical thinking, and support understanding of both the ecological problem and the processes underlying each research study.
  • Thumbnail Image
    Item type:Item,
    A espiral do cuidado: reflexões sobre o autocuidado na minha formação médica
    (Universidade Federal de São Carlos, 2026-10-27)
    Brito, Thaís de
    ;
    Adler, Maristela Schiabel
    ;
    http://lattes.cnpq.br/3896827914383636
    ;
    https://orcid.org/0000-0003-0222-9856
    ;
    branca
    This work presents a critical-reflective memoir of the author's trajectory through medical school at the Federal University of São Carlos, from 2021 to 2026, through the lens of self-care. Drawing on the ethical commitment, incorporated into the Declaration of Geneva, that physicians must attend to their own health and well-being, and on the high prevalence of psychological distress among medical students in Brazil, the text revisits experiences from the basic cycle, the clinical cycle and the internship, in dialogue with the literature on mental health in medical education and physician well-being. The narrative follows moments of overload and emotional distress, as well as the bonds and achievements that marked each stage. Using as a metaphor the constructivist spiral that structures the program's curriculum, the reflection shows that self-care, like medical learning, is built in layers, lost and regained throughout training, and sustained by a support network. It concludes that caring for oneself is a precondition for caring for others and an inseparable part of medical identity, and that promoting self-care should also be a responsibility of medical education.
  • Thumbnail Image
    Item type:Item,
    Elaboração e análise de indicadores de patentes sobre produção de pó e ligas de alumínio utilizadas em manufatura aditiva
    (Universidade Federal de São Carlos, 2026-09-02)
    Andrade, Luis Andre Sousa
    ;
    Faria, Leandro Innocentini Lopes de
    ;
    http://lattes.cnpq.br/0767710394930118
    ;
    https://orcid.org/0000-0002-8369-1315
    ;
    Gargarella, Piter
    Aluminum (Al) alloys for additive manufacturing (AM) present challenges such as high reflectivity, susceptibility to solidification cracking, and defect formation, which have been gradually overcome through the development of new alloys and the optimization of powder production. This study aimed to develop and analyze patent indicators on powder production and aluminum alloy development for AM, mapping the state of the art, technological trends, and key players in the field, based on data from the Derwent Innovations Index (DII) database processed with Natural Language Processing (NLP) tools in VantagePoint software. The results reveal a gradual shift from the use of conventional aluminum alloys, particularly those belonging to the 6xxx and 7xxx series, toward compositions specifically designed for AM. Modified alloys containing Ti, Zr, and Sc are particularly prominent, primarily because of their potential to reduce solidification cracking susceptibility and control defect formation during processing. Regarding powder production, gas atomization remains the dominant technology due to its high productivity, scalability, and cost competitiveness. Nevertheless, alternative approaches, including ultrasonic and centrifugal atomization, continue to attract technological interest because of their ability to produce powders with improved morphological characteristics and reduced defect populations. The geographical analysis identifies China as the leading center of patent activity, with universities and research institutions playing a major role in the development of new alloy compositions, whereas companies are more strongly focused on gas atomization technologies. The resulting indicators provide strategic support for research, development, and intellectual property protection in the sector, enabling the identification of innovation gaps and the monitoring of global competition.
  • Thumbnail Image
    Item type:Item,
    Efeitos do ensino de ecoico sobre emergência de tato com unidades verbais complexas em crianças com TEA
    (Universidade Federal de São Carlos, 2026-08-21)
    Yamada, Viviane Macedo
    ;
    Goyos, Antonio Celso de Noronha
    ;
    https://lattes.cnpq.br/3407008711535651
    ;
    https://orcid.org/0000-0002-6150-1074
    ;
    https://lattes.cnpq.br/0512314068392954
    Os operantes verbais são definidos e classificados a partir das relações funcionais estabelecidas entre respostas e as condições de estímulos antecedentes e consequentes que as controlam. No contexto do Transtorno do Espectro Autista (TEA), o ensino sistemático desses operantes tem sido amplamente recomendado como estratégia fundamental para o desenvolvimento da linguagem funcional, contribuindo para a ampliação das habilidades comunicativas e da autonomia. A literatura na área da Análise do Comportamento Aplicada indica, de modo consistente, que os operantes verbais tendem a apresentar independência funcional, de modo que o ensino direto de um operante não garante, necessariamente, a emergência espontânea de outros. Contudo, investigações mais recentes têm demonstrado que, sob condições específicas de ensino, especialmente aquelas que envolvem múltiplos exemplares e o fortalecimento de repertórios relacionados, pode ocorrer a emergência de novos operantes não diretamente ensinados. Diante desse cenário, o presente estudo buscou ensinar respostas ecoicas com mais de uma palavra, com a finalidade de avaliar a emergência do operante verbal tato com múltiplas palavras. Adotou-se um delineamento experimental de linha de base múltipla entre participantes, adequado para a análise de relações funcionais em contextos aplicados. A variável independente consistiu na implementação de um procedimento de ensino envolvendo treino de ecoico estruturado. A variável dependente correspondeu à emissão de respostas de tato com mais de uma palavra diante de estímulos visuais apresentados. Os resultados indicaram a emergência de respostas de tato com duas e três palavras após a implementação do procedimento de ensino, sem ensino direto desse operante. Entretanto, o desempenho variou entre os participantes e as condições experimentais, e nem sempre atingiu o critério de acerto estabelecido. Também foram observadas mudanças graduais na extensão das respostas e indícios de ampliação do controle discriminativo por múltiplas propriedades dos estímulos. Foram observadas alterações na topografia das respostas emitidas pelos participantes, as quais passaram a ocorrer sob controle de múltiplos elementos de um único estímulo discriminativo, sugerindo uma ampliação no controle de estímulos.