Início » Tecnologia & Inovação » Webinars Técnico » 2020
10/09 - 11:00 am
Petar Zecevic
Abstract ⓘ
Abstract
×Speaker
Petar Zecevic
Title
Using AXS for large-scale astronomical analyses (University of Zagreb)
Abstract
AXS is a distributed system based on Apache Spark used for fast cross-matching of astronomical catalogs and for doing general analyses of astronomical data. In this talk, Petar will give a short introduction to Apache Spark and AXS and delve into inner workings of AXS and explain what makes it so fast. He will also show a live demo with several examples.
Slides not available
01/09 - 01:30 pm
Luis F. R. Macedo
Abstract ⓘ
Abstract
×Speaker
Luis F. R. Macedo
Title
Bancos de dados para grandes volumes (VMware Tanzu)
Abstract
Análise de dados com grandes volumes de informação sempre foi um desafio computacional. Quando temos um volume de dados acima do que um computador consegue trabalhar de forma razoável temos que partir para soluções de computação distribuída que muitas vezes são complexas e pouco eficientes. O Greenplum Database é um banco de dados paralelo baseado em PostgreSQL que pode administrar petabytes de dados sem a necessidade de hardwares proprietários e com uma familiar interface SQL.
25/08 - 01:00 pm
Fabio Porto
Abstract ⓘ
Abstract
×Speaker
Fabio Porto
Title
Introdução ao Ecosistema e Modelo de Programação no Apache Spark (Laboratório Nacional de Computação Científica)
Abstract
O processamento de grandes volumes de dados passou a ser a norma no suporte a pesquisas in-silico, como acontece com a eAstronomia. Neste webminar, discutiremos o framework Apache Spark que já vem se tornando uma referência de processamento Big Data. Apache Spark é integrado ao sistema de arquivos distribuídos HDFS para oferecer um ambiente de execução baseada em localidade de dados, com escalabilidade para grande número de nós de computação, e mecanismos de tolerância à falhas. O framework adota o modelo de processamento em memória reduzindo significativamente o custo evolvido em tarefas envolvendo iterações, muito utilizadas em processos baseados em aprendizagem.
Slides not available
11/08 - 11:00 am
Kyle Chard
Abstract ⓘ
Abstract
×Speaker
Kyle Chard
Title
funcX: a federated function serving fabric for science (University of Chicago)
Abstract
Exploding data volumes and velocities, new computational methods and platforms, and ubiquitous connectivity demand new approaches to computation in the sciences. These new approaches must enable computation to be mobile, so that, for example, it can occur near data, be triggered by events (e.g., arrival of new data), be offloaded to specialized accelerators, or run remotely where resources are available. They also require new design approaches in which monolithic applications can be decomposed into smaller components, that may in turn be executed separately and on the most suitable resources. To address these needs we present funcX—a distributed function as a service (FaaS) platform that enables flexible, scalable, and high performance remote function execution. funcX’s endpoint software can transform existing clouds, clusters, and supercomputers into function serving systems, while funcX’s cloud-hosted service provides transparent, secure, and reliable function execution across a federated ecosystem of endpoints. We motivate the need for funcX with several scientific case studies, present our prototype design and implementation, show optimizations that deliver throughput in excess of 1 million functions per second, and demonstrate, via experiments on two supercomputers, that funcX can scale to more than more than 130,000 concurrent workers.
Slides not available
23/07 - 11:00 am
Michelle Wangham
Abstract ⓘ
Abstract
×Speaker
Michelle Wangham
Title
Diretrizes e Boas Práticas para Adequação à LGPD (Guidelines and Best Practices for LGPD compliance) (Universidade do Vale do Itajaí)
Abstract
A Lei Geral de Proteção de Dados (LGPD) altera significativamente o marco normativo referente à proteção dos dados pessoais mantidos por todas as organizações (empresariais e públicas). As corporações que já começaram sua jornada de conformidade têm encontrado inúmeras dificuldades para a construção de suas diretrizes de tratamento de dados pessoais e de políticas para prevenção e mitigação em caso de incidentes. A partir de uma abordagem interativa, analisarei nesta palestra as etapas para adequação à LGPD e as principais dores e desafios da jornada da conformidade. Abordarei um conjunto de diretrizes e boas práticas para proteção de dados que aliem técnicas de segurança cibernética, segurança jurídica e governança corporativa. -- The Brazilian General Data Protection Law significantly changes the regulatory milestone related to personal data protection, adopted by all public and private organizations. The organizations that have already initiated this compliance journey have come across many difficulties in building their guidelines for personal data treatment and preventing and mitigating security incidents. I shall analyze the steps for LGPD compliance and its main challenges from an interactive approach. I shall also discuss a group of best practices to protect personal data that align cybersecurity controls, legal security, and corporate governance. (This talk is in Portuguese)
09/07 - 11:00 am
François Lanusse
Abstract ⓘ
Abstract
×Speaker
François Lanusse
Title
Merging deep learning with physical models for the analysis of modern cosmological surveys (CEA Paris-Saclay)
Abstract
The upcoming generation of cosmological surveys such as LSST will aim to map the Universe in great detail and on an unprecedented scale. This of course implies new and outstanding challenges at all levels of the scientific analysis. In this talk, I will illustrate how recent advances in Deep Learning, and associated automatic differentiation tools (i.e. TensorFlow), can help us tackle these challenges and rethink our approach to data analysis for cosmological surveys. At the image level, combining physical models of the instrument (which account for noise/PSF) with deep generative models (which account for complex galaxy morphologies) can allow us to solve a range of astronomical inverse problems ranging from deconvolution to deblending galaxy images. At the cosmological analysis level, I will present our efforts to implement N-body simulations directly in TensorFlow, opening the door to a range of novel and efficient inference techniques.
04/06 - 11:00 am
Carlos Alexandre Wuensche
Abstract ⓘ
Abstract
×Speaker
Carlos Alexandre Wuensche
Title
21 cm cosmology and the BINGO radio telescope (INPE)
Abstract
Cosmology in the XXI century is experiencing a \\\\\\\\\\\\\\\"Golden Age\\\\\\\\\\\\\\\", with observations and theoretical models contributing to a large-scale description of the Universe. The current view is that it can be well described by the so-called Lambda-CDM model, but some open problems challenge physics and cosmology, including the origin and properties of so-called dark energy. The so-called baryonic acoustic oscillations (BAO), detected for the first time in 2005, are considered one of the most effective probes to understand the properties of dark energy. However, given the implications of these measurements, it is important that they are confirmed at other wavelengths and measured over a wide range of redshifts. The radio band provides a unique and complementary observation window, by emitting 21 cm of neutral hydrogen. The redshifted 21 cm (1420 MHz) emission of the hyperfine transition of neutral hydrogen is measured at lower frequencies, so that the observation frequency is converted directly into information about the source\\\\\\\\\\\\\\\'s redshift. The BINGO radio telescope (BAO from Integrated Neutral Gas Observations) is a new instrument, designed specifically to observe BAO, mapping a redshift band between 0.13 and 0.45. This seminar will present the basics of 21 cm BAO cosmology, the intensity mapping technique used and describe the current development status of the BINGO radio telescope.
28/05 - 04:00 pm
Mario Juric
Abstract ⓘ
Abstract
×Speaker
Mario Juric
Title
Minimoons to Planet X: Mapping Solar System Populations with ZTF and LSST (University of Washington/LSST)
Abstract
The small bodies of the Solar System are a valuable tracer of its present-day structure, its evolution, and ultimately provide clues into the early times of its formation. In the next 5 years, the known sample of all small body populations will grow 10-50x, driven largely by large survey programs. For example, the LSST alone will generate a billion measurements of millions of Solar System objects, with simulations predicting ~100,000 new discoveries of nearby NEOs (Jones et al. 2017), 5.5 million for the main belt, and ~40,000-200,000 for the trans-Neptunian populations (Ivezic et al 2008; Juric et al. 2018). In this talk I will discuss what to expect from this sample, especially in the early years. I\\\\\\\'ll briefly overview the LSST and ZTF surveys, and what they are expected to discover (and -- in the case of ZTF -- already have discovered). The talk will discuss the techniques to find small bodies in survey datasets, and places where improved algorithms could significantly increase the yields. I will conclude with an overview the science opportunities this new sample is bringing.
Slides not available
14/05 - 11:00 am
Emille Ishida
Abstract ⓘ
Abstract
×Speaker
Emille Ishida
Title
Machine Learning in Astronomy (Université Clermont Auvergne)
Abstract
The availability of large data sets revolutionized many areas of scientific research, astronomy included. The current -- and in many ways already overwhelming -- data paradigm will suffer still another revolution with the advent of the new generation of large astronomical surveys. In this new scenario, the use of automated methods of analysis will be unavoidable. In this talk, I will give a short introduction to the basic principles of machine learning and describe situations where they are traditionally used in astronomical research. I will also present how domain knowledge can be used to optimize results from traditional algorithms by incorporating expert feedback in the learning process.
Slides not available
05/03 - 03:00 pm
Joseph Masiero
Abstract ⓘ
Abstract
×Speaker
Joseph Masiero
Title
NEOWISE tools and techniques (JPL NASA)
Abstract
The NEOWISE infrared space telescope has provided an unprecedented set of infrared photometry and astrometry of the entire sky. This includes stars and galaxies, as well as over 150,000 asteroids and comets in our Solar system. I will provide an overview of the data that are publicly available, the tools and techniques that are included to access this data, and some of the results that have been derived from these measurements.