O Ciclo de Seminários PESC tem como objetivo trazer palestras acessíveis a um público mais amplo
ministradas por pesquisadores e professores mais experientes. As palestras ao longo do ano terão
tema e foco variados podendo ser mais específicas (ex. avanços no contexto de um problema específico)
ou mais abrangentes (ex. desafios de uma área).
A apresentação e discussão de ideias novas e antigas de diferentes temas contribui de
maneira fundamental para a formação e pesquisa desenvolvida por alunos e professores.
As palestras do Ciclo de Seminários PESC ocorrem em geral uma vez ao mês, mas sempre
quarta-feira às 11h na sala H-324B. Segue a programação confirmada para 2016. Não percam!
|
Exploring Heterogeneity within a Core for Improved Power Efficiency
Sandip Kundu, Full Professor, University of Massachusetts (UMass), EUA
16 de março
Resumo - Slides
Asymmetric multi-core processors (AMPs) comprise of cores with different sizes of micro-
architectural resources yielding very different performance and energy characteristics. Since the
computational demands of workloads vary from one task to the other, AMPs often provide
greater power efficiency than symmetric multicores. Furthermore, as the computational demands
of a task change during its course of execution, reassigning the task from one core to another,
where it can run more efficiently can further improve the overall energy efficiency. However, too
frequent re-assignments of tasks to cores may result in high overhead. To greatly reduce this
overhead we propose a morphable core architecture that dynamically adapts its resource sizes
and operating frequency to assume one of four possible core configurations. Such a morphable
architecture allows more frequent task to core configuration re-assignments for a better match
between the current needs of the task and the available resources. To make the online morphing
decisions we have developed a runtime analysis scheme using hardware performance counters.
Our results indicate that the proposed morphable architecture controlled by the runtime
management scheme can improve the performance/watt of applications by 43% over executing
on a static AMP.
Biografia resumida
Sandip Kundu is a Professor at the University of Massachusetts at Amherst. Prior to joining
academia, he spent several years in industry: first as a Research Staff Member at IBM Research
Division and then at Intel Corporation as a Principal Engineer. He has published well-over 200
research papers in VLSI design and test and holds several key patents including ultra-drowsy
sleep mode in processors, and has given more than a dozen tutorials at various conferences. He is
a Fellow of the IEEE, Fellow of the Japan Society for Promotion of Science (JSPS), Senior
International Scientist of the Chinese Academy of Sciences and a Distinguished Visitor of the
IEEE Computer Society. He is currently an Associate Editor of the IEEE Transactions on
Dependable and Secure Computing. Previously, he has served as an Associate Editor of the IEEE
Transactions on Computers, IEEE Transactions on VLSI Systems and ACM Transactions on
Design Automation of Electronic Systems. He has been Technical Program Chair/General Chair
of multiple conferences including ICCD, ATS, ISVLSI, DFTS and VLSI Design Conference.
|
|
Entering the Information Age
John Hopcroft, IBM Professor of Engineering and Applied Mathematics, Cornell University, EUA
3 de maio (terça às 15h no auditório da COPPE no CT2)
Website da visita
Cartaz - Resumo - Slides
The world is entering the Information Age which is changing the education needed to prepare for good
jobs and successful future careers. We are now concerned with extracting information from the
enormous data sets that are available. Information, such as medical records, previously stored on paper
will be digitized raising issues about how to preserve privacy. New ideas, such as zero knowledge
proofs, are being developed. This talk will discuss the Information Age and give examples of the
mathematics needed to undergird the education necessary for the jobs of the future.
Biografia resumida
John E. Hopcroft is the IBM Professor of Engineering and Applied Mathematics in Computer Science at Cornell University. From January 1994 until June 2001, he was the Joseph Silbert Dean of Engineering. After receiving both his M.S. (1962) and Ph.D. (1964) in electrical engineering from Stanford University, he spent three years on the faculty of Princeton University. He joined the Cornell faculty in 1967, was named professor in 1972 and the Joseph C. Ford Professor of Computer Science in 1985. He served as chairman of the Department of Computer Science from 1987 to 1992 and was the associate dean for college affairs in 1993. An undergraduate alumnus of Seattle University, Hopcroft was honored with a Doctor of Humanities Degree, Honoris Causa, in 1990.
Hopcroft's research centers on theoretical aspects of computing, especially analysis of algorithms, automata theory, and graph algorithms. He has coauthored four books on formal languages and algorithms with Jeffrey D. Ullman and Alfred V. Aho. His most recent work is on the study of information capture and access.
He was honored with the A. M. Turing Award in 1986. He is a member of the National Academy of Sciences (NAS), the National Academy of Engineering (NAE) and a fellow of the American Academy of Arts and Sciences (AAAS), the American Association for the Advancement of Science, the Institute of Electrical and Electronics Engineers (IEEE), and the Association of Computing Machinery (ACM). In 1992, he was appointed by President Bush to the National Science Board (NSB), which oversees the National Science Foundation (NSF), and served through May 1998. From 1995-98, Hopcroft served on the National Research Council's Commission on Physical Sciences, Mathematics, and Applications.
|
|
Intratabilidade e Otimização
Celina M. H. de Figueiredo, Professor Titular, PESC/COPPE/UFRJ
Luciana S. Buriol, Professor Associado, II/UFRGS
Eduardo Uchoa Barboza, Professor Associado, EP/UFF
22 de junho (às 13h30m) - em conjunto com Seminário de Grafos e Algoritmos
Resumo - Slides 1 - Slides 2
Faremos uma homenagem a David Johnson (1945-2016), destacando
as suas contribuições para a análise teórica e experimental de
algoritmos. Ao longo da sua brilhante carreira de 40 anos no Bell Labs
Research, foi chefe do departamento de Fundamentos Matemáticos de
Computação e do departamento de Algoritmos e Otimização. David Johnson
liderou na ACM a área de Algoritmos e Teoria da Computação, através da criação da
conferência ACM-SIAM SODA e do grupo de interesse ACM SIGACT. O seu
livro "Computers and Intractability: A Guide to the Theory of
NP-Completeness" e a sua série "An Ongoing Guide on NP-completeness"
constituem os fundamentos para o desenvolvimento da teoria que
identifica os problemas difíceis. Ele criou e liderou nos últimos 25
anos as DIMACS Implementation Challenges para computação experimental
buscando o rigor científico na avaliação empírica de algoritmos.
Biografia resumida
Celina Figueiredo obteve seu doutorado há 25 anos no Programa de
Engenharia de Sistemas e Computação da COPPE sob a orientação de Jayme
Szwarcfiter. Fez carreira docente na UFRJ, onde ingressou no Instituto
de Matemática em 1989, e na COPPE em 1991. Fez pós-doutorado em 1995
na Universidade de Waterloo, no Canadá. É professora titular do PESC,
onde coordena desde 2010 o Núcleo de Excelência em Algoritmos
Randomizados, Quânticos, e Aproximativos: Projeto, Análise e
Implementação de Soluções Eficientes para problemas Combinatórios
Fundamentais. É desde 2005 Cientista do Nosso Estado FAPERJ. Recebeu
em 2006 o Prêmio Giulio Massarani de Mérito Acadêmico da COPPE e em
2013 homenagem na solenidade comemorativa dos 50 anos da COPPE.
Recentemente, foi paraninfa da turma de Graduação em Engenharia de
Computação e Informação da Escola Politécnica.
Eduardo Uchoa é professor do Departamento de Engenharia de Produção da
UFF. Sua principal área de pesquisa é o uso de programação inteira
para a resolução prática de problemas de otimização combinatória
NP-difíceis, tais como os problemas de roteamento de veículos ou o
problema de Steiner. Em particular, é um especialista no uso conjunto
das técnicas de geração de colunas e separação de cortes, nos chamados
algoritmos de branch-cut-and-price.
Luciana Buriol possui doutorado em otimização pela FEEC/UNICAMP, com
doutorado sanduíche na AT&T Labs Research, USA. Possui pos-doutorado
em algoritmos pela Universidade de Roma La Sapienza, Itália. Desde
2006 é professora do Instituto de Informática da Universidade Federal
do Rio Grande do Sul. Atua nas áreas de otimização combinatória e
algoritmos. Foi Presidente da ALIO - Asociación Latino-Iberoamericana
de Investigación Operativa na gestão 2012-2014, e atualmente é
Vice-Presidente da IFORS - International Federation of Operational
Research Societies, na gestão 2016-2018.
|
|
As Contribuições Científicas de Diffie e Hellman - Prêmio Turing 2015
Luis Menasché Schechter, Professor Adjunto, DCC/UFRJ
20 de julho (às 13h30m) - em conjunto com Seminário de Grafos e Algoritmos
Resumo - Slides - ACM Turing Award Diffie and Hellman Video
O Prêmio Turing é um prêmio concedido anualmente para cientistas da computação, sendo considerado o equivalente do Prêmio Nobel para Computação. Os mais recentes vencedores do prêmio Turing são Whitfield Diffie e Martin Hellman, pesquisadores da área de criptografia. Nesta palestra, tentaremos oferecer uma visão panorâmica sobre as principais contribuições destes dois pesquisadores para o avanço desta área de pesquisa, destacando a revolução que causaram ao inaugurarem duas novas áreas de pesquisa: a criptografia de chave pública e as assinaturas digitais.
Biografia resumida
O autor trabalha na área de Criptografia de Chave Pública, estudando problemas matemáticos que proporcionam a segurança subjacente aos diversos métodos e também as questões relativas à complexidade computacional dos algoritmos envolvidos. O autor também possui grande interesse na pesquisa da História da Ciência da Computação e na sua divulgação para o público geral. Nos últimos anos, tem se dedicado ao estudo e divulgação da vida e das contribuições científicas do matemático inglês Alan Turing (1912-1954).
|
|
The Versatility of TTL Caches for cache service differentiation and pricing
Don Towsley, Distinguished University Professor, University of Massachusetts (UMass), Amherst, EUA
14 de setembro às 11h
Resumo - Slides
In this talk we review the versatility of timer driven caches, also known as TTL (time to live) caches. TTL caches are extremely useful, both for designing new cache management policies that can provide differentiated services and as models for classical and new cache replacement policies such as LRU, random, and first in first out. The first part of the talk focuses on the use of TTL caches for providing service differentiation among contents. This relies on the assignment of utilities to content and the use of the utility maximization framework to develop on-line algorithms to maximize the sum of contnet utilities. One of the byproducts of this work is the reverse engineering of classical cache replacement policies as solutions to this cache utilization optimization problem and the identification of the utility functions underlying these policies.
In the second part of the talk we focus on the problem of how cache providers can and should incentivize the services that they provide content providers. We examine two classes of mechanisms that differ according to whether the cache provider charges for the time that the content occupies the cache or the content request rate. In both cases the cache is a TTL cache and the problem that ensues is that of selecting appropriate values for the timers. The conclusion is that a policy that charges according to request rate the latter policy is better suited for providing incentives to the cache provider.
Biografia resumida
Don Towsley holds a B.A. in Physics (1971) and a Ph.D. in Computer Science (1975) from University of Texas. He is currently a Distinguished Professor at the University of Massachusetts in the College of Information & Computer Sciences. He has held visiting positions at numerous universities and research labs. His research interests include networks and performance evaluation.
He is co-founder and Co-EiC of the new ACM Transactions on Modeling and Performance Evaluation of COmputing Systems (TOMPECS), and has served as Editor-in-Chief of IEEE/ACM Transactions on Networking and on numerous editorial boards. He has served as Program Co-chair of several conferences including INFOCOM 2009.
He is a corresponding member of the Brazilian Academy of Sciences and has received several achievement awards including the 2007 IEEE Koji Kobayashi Award and the 2011 INFOCOM Achievement Award. He has received numerous paper awards including the 2012 ACM SIGMETRICS Test-of-Time Award, a 2008 SIGCOMM Test-of-Time Paper Award, and 2015 and 2016 ACM SIGMETRICS Best Paper Awards. Last, he has been elected Fellow of both the ACM and IEEE.
|
|
Global and local stability of multi-dimensional Markov chains
Seva Shneer, Associate Professor, Heriot-Watt University, Escócia
26 de outubro às 11h
Resumo - Slides
We will look at multi-dimensional Markov chains where the components may be dependent but one of them has a predictable behaviour (is, in some sense, stable or unstable). We will be focused on finding conditions sufficient for the other components (or the entire chain) to have a limiting distribution. Processes with the properties described in this talk appear naturally in a number of applications and we will discuss examples of these applications.
Biografia resumida
Vsevolod (Seva) Shneer received a PhD Degree from Heriot-Watt University in 2007 and then worked as a postdoc reseacher t EURANDOM, the Netherlands, and at EPFL, Switzerland. He moved back to Heriot-Watt University in 2010 and now holds an Associate Professor position.
|
|
Dealing with Discriminatory Data Mining
Christopher W. Clifton, Professor, Purdue University, EUA
7 de novembro às 11h
Resumo - Slides
There is growing evidence that algorithms running on "Big Data"
can lead to outcomes that are biased against underrepresented
groups. This is in spite of the fact that such group information
(race, gender, religion, etc.) is not used by the algorithms.
This talk will discuss some of the issues, pointing out evidence,
and hypothesize causes. We will then look at one solution,
based on adapting a Bayesian network classifier to reduce
disparate impact on groups that are treated "differently" by
the originally learned classifier.
This talk is based on work with Koray Mancuhan that appeared in
Artificial Intelligence and Law (2014) 22:211-238.
Biografia resumida
Dr. Clifton is a Professor of Computer Science at Purdue University.
He works on data privacy, particularly with respect to analysis of
private data. From 2013-2015, Dr. Clifton served as a program
director at the National Science Foundation. Prior to joining Purdue
in 2001, he was a principal scientist in the Information Technology
Division at the MITRE Corporation. Before joining MITRE in 1995, he
was an assistant professor of computer science at Northwestern
University.
|
Entre em contato e envie seus comentários e sugestões, inclusive de potenciais palestrantes.
Organizado por Daniel R. Figueiredo.
Anos anteriores:
2014,
2015
