Horário: 14h
Palestrante: Léia Sousa de Sousa (doutorado)
Orientador: Prof Andre Drummond
Título: Engenharia de Tráfego em Redes Ópticas Metropolitanas
Resumo: Com os crescentes requisitos de largura de banda para as pessoas e o desenvolvimento da urbanização, o movimento da população na cidade tem uma influência crescente na distribuição do tráfego nas dimensões de espaço e tempo. A distribuição desequilibrada resulta no bloqueio de serviços em diferentes áreas e em diferentes períodos de tempo e reduz a utilização de recursos de espectro em redes ópticas elásticas. Esta pesquisa propõe um algoritmo de alocação que foca na redução do bloqueio geral e por área.
Horário: 14h20
Palestrante: Liriam Enamoto (doutorado)
Orientador: Prof Li Weigang
Title: Generic Framework for Multimodal Few-shot Learning
Abstract: to be defined
Horário: 14h40
Palestrante: Lucas Silva Lopes (doutorado)
Orientador: Prof Ricardo Queiroz
Title: Context modeling for arithmetic coding
Abstract: Our objective is to discuss the use of neural networks in context modeling for arithmetic coding, compared to other more traditional methods such as those based on look up tables. We delve into the theory, and present numerical results to support the theory.
Horário: 15h
Palestrante: Guilherme Enéas Vaz Silva (doutorado)
Orientador: Prof Andre Drummond
Title: Deep-Quality-EON Classifier and DRL approach
Abstract: to be defined
Horário: 15h20
Palestrante: Patricia Medyna Lauritzen de Lucena Drumond (doutorado)
Orientador: Prof Teófilo Campos
Title: Visual and Textual Feature Fusion for Document Analysis
Abstract: The large volume of documents produced daily in all sectors, such as industry, commerce, and government agencies, has increased the number of researchs aimed at automating the process of reading, understanding, and analyzing documents. Business documents can be born digital, as electronic files, or can be a digitized form that comes from writing or printed on paper. In addition, these documents often come in various layouts and formats. They can be organized in different ways, from plain text, multi-column layouts, and a wide variety of tables/forms/figures. In many documents, the spatial relationship of text blocks usually contains important semantic information for downstream tasks. The relative position of text blocks plays a crucial role in document understanding. However, the task of embedding layout information in the representation of a page instance is not trivial. In the last decade, Computer Vision and Natural Language Processing pre-training techniques have been advancing in extracting content from document images considering visual, textual, and layout features. Deep learning methods, especially the pre-training technique, represented by Transformer architecture, have become a new paradigm for solving various downstream tasks. However, a major drawback of such pre-trained models is that they require a high computational cost. Unlike these models, we propose a simple and traditional rule-based spatial layout encoding method, which combines textual and spatial information from text blocks. We show that this enables a standard NLP pipeline to be significantly enhanced without requiring expensive mid or high-level multimodal fusion. We evaluate our method on two datasets, Tobacco800 and RVL-CDIP, for document image classification tasks. The document classification performed with our method obtained an accuracy of 83.6% on the large-scale RVL-CDIP and 99.5% on the Tobacco800 datasets. In order to validate the effectiveness of our method, we intend to carry out more experiments. First, we will use other more robust datasets. Then we will change parameters such as quadrant amounts, insertion/deletion of positional tokens, and other classifiers.
Horário: 15h40
Palestrante: Hércules de Campos Junior (mestrado)
Orientadora: Profa Priscila Sollis
Título: Um estudo de distribuição de tráfego e aspectos de segurança em Open RAN
Resumo: a definir
Local: Teams- Equipe PPGI-316415 Seminário, Canal 2-2022
https://teams.microsoft.com/l/channel/19%3a05900df7390e45edaa77283171cbb44b%40thread.tacv2/Semin%25C3%25A1rios%25202-2022?groupId=93b66213-b249-467a-bcbe-dcd4255edf95&tenantId=ec359ba1-630b-4d2b-b833-c8e6d48f8059
Profa Célia Ghedini Ralha (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo.)
Coordenadora Seminários de Pós-Graduação em Informática 2-2022