• Improving approximate nearest neighbor search in HNSW graphs with data clustering 

  Vieira, Marco Antônio Athayde de Aguiar (2025) [Dissertação]

  Exhaustive strategies for search and information retrieval tasks over vector databases do not meet the runtime efficiency expectations of modern applications. Approximate Near- est Neighbor Search (ANNS) algorithms use ...
• Recursive filtering techniques for edge-aware filter analysis and implementation 

  Ferreira, Hermes Hofmeister (2025) [Tese]

  In this work, we aim to develop a formal mathematical method of analyzing edgeaware digital filters similar to how Fourier analysis can be used to understand linear filters. We believe there is a need for such a theory, ...
• Zero-shot human attribute classification using pretrained VLMs 

  Ceschini, Lucas Mayer (2025) [Dissertação]

  Gender and ethnicity recognition is a key computer vision task in biometrics, surveil- lance, and demographic studies. Contrastive Language-Image Pre-training (CLIP) is a Vision-Language Model with promising zero-shot ...
• Approximate logic synthesis combining AIG and mapped flows 

  Pinho, Gabriel Ammes (2025) [Tese]

  As digital systems grow in complexity, technology scaling driven by Moore’s Law is approaching its physical and economic limits. Many widely used applications, however, exhibit error resilience, allowing acceptable results ...
• Reactive and proactive approaches to predict and identify critical flows using meta-heuristic in programmable data planes 

  Silva, Marcus Vinicius Brito da (2025) [Tese]

  Predicting critical flows that can potentially harm network performance is a challenging task. As such, network operators need to constantly monitor, analyze, and configure network devices to allow users to better experience ...
• A vector-based representation of dispositions 

  Antunes, Cauã Roca (2025) [Tese]

  Dispositions are properties that only manifest under certain conditions. Examples include the fragility of a glass or the flammability of a match. These properties may exist in a latent state until the proper conditions ...
• iUXRV : um questionário padronizado em português brasileiro para avaliação de experiência de usuário em realidade virtual imersiva 

  Cheiran, Jean Felipe Patikowski (2025) [Tese]

  Conforme a Realidade Virtual (RV) imersiva vem se popularizando por meio de dispositivos mais baratos e de melhor qualidade, profissionais e pesquisadores adotam diferentes estratégias de avaliação para essas tecnologias. ...
• LLMusic: Modelagem de tópicos em letras de músicas combinando LLM, Engenharia de Prompt e BERTopic - Estudo de Caso no Funk Brasileiro 

  Rojas, Jesus Daniel Yepez (2025) [Dissertação]

  Letras de músicas impõem desafios à modelagem de tópicos pela presença de linguagem figurada, gírias e discursos implícitos. Este trabalho propõe o LLMusic, framework que integra LLMs, Engenharia de Prompt (EP) e BERTopic. ...
• An approach based on language models and natural language definitions for relationship classification 

  Amorim, Marina Martins (2024) [Dissertação]

  Identifying relationships between concepts is a very important task for several NLP tasks, as well as for building explicit knowledge models (ontologies and knowledge graphs). In many of these tasks, experts usually manually ...
• IoTP4Chain : explorando planos de dados programáveis para forense eficiente em IoT utilizando blockchain 

  Susin, Robert Patrick (2025) [Dissertação]

  A forense em Internet das Coisas (IoT) encara desafios como aquisição, identificação e análise de evidências. Esse processo é complexo, por haver um crescimento exponencial no número de dispositivos IoT e pelos requisitos ...

Ver mais