Local: Sala Multiuso CIC
Horário: 14h
Palestrante: Daniel Saad Nogueira Nunes (doutorado)
Título: A Grammar Compression Algorithm based on Induced Suffix Sorting
Resumo: This work presents a new grammar compression algorithm (GC-IS) based on the suffix sorting algorithm of Nong et al. In the original algorithm, by classifying the suffixes in L-Type, S-Type and LMS-Type categories, it is possible to induce the relative order of the LMS-Substrings which are replaced for integer factors and used for obtaining the solution to the recursive problem in order to acquire the suffix array. By using such factors to encode the LMS-Substrings and applying proper compression techniques, it is possible to achieve good compression ratios. Regarding repetitive data, and the popular compressors 7zip and gzip, it was shown that GC-IS has a competitive compression ratio and encoding time and offers an alternative option to compress such data.
Horário: 14h30
Palestrante: Ariane Alves Almeida (doutorado)
Título: Formalização de Terminação por Pares Dependentes para Programas Funcionais de Primeira Ordem
Resumo: O uso de pares dependentes para provar terminação foi inicialmente definido para o contexto de Sistemas de Reescrita de Termos, mas que pode também ser trazido para o contexto de programas funcionais. A partir disso pode-se formalizar este critério, bem como sua equivalência com outros critérios para este contexto, como Grafos de Contexto de Chamadas e Grafos com Medidas de Peso, ou mesmo com definições semânticas de terminação. Essas equivalências são formalizadas no assistente de prova usando a linguagem PVS0, onde deseja-se então inserir essa nova noção de terminação e também provar sua equivalência.
Horário: 15h
Palestrante: Eneias Cordeiro da Silva (mestrado)
Título: A Domain Specific Modeling Approach for Automation of Scientific Experiments
Resumo: Experimentation is a means to produce controlled changes and to measure the variables involved in the phenomena under study. In addition, it provides data to further replication. However, conducting an experiment is a complex and time-consuming task. This complexity is observed in all phases of the experimentation process. Furthermore, validity and reproducibility must be addressed from the earliest phases of the process to achieve a reproducible experiment and valid results. To address these issues, many techniques have been proposed. Although these techniques help in conducting controlled experiments, none of them simultaneously addresses specification of the experiments, automated treatment execution, and automated data analysis. Thus, we propose a Domain Specific Modeling (DSM) approach for automation of scientific experiments. The solution will provide means to the specification of the experiments, automated execution, and utomated analysis. The solution will be evaluated with respect to validity, reproducibility, time consumption, usability, and self-configuration. We expect to reduce time consumption and improve the validity of the experiments. We also expect to improve the rigor and the reproducibility of the experiments.
Horário: 15h30
Palestrante: Helard Alberto Becerra Martinez (doutorado)
Título: Subjective experiment results using an immersive methodology
Resumo: In this work, we present the results of a set of psychophysical experiments in which a group of volunteers rated the quality of a set of audio-visual sequences. The sequences had up to three types of distortions: video coding, packet-loss, and frame freezing distortions for the video component. Additionally, four types of audio artifacts were inserted on the sequences: 1) background noise, 2) chopped speech, 3) clipped speech, and 4) echo speech. The experiments were conducted using the basic directions of the immersive experimental methodology.
Organizadora: Profa Célia Ghedini Ralha (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo.)
Coordenadora dos Seminários de Pós-Graduação Informática 2017-1