29 maio 2015
Links
Acionista diz que EY sabia da corrupção da WalMart no MéxicoNovo livro de Richard Thaler: a história das finanças comportamentais
Novo livro de Dubner: Quando roubar um banco
Quando é melhor beber café? (vídeo)
ECT faz reversão de provisão contábil e consegue obter lucro
Supremo exige a evidenciação da operação de empréstimo do BNDES ao JBS
28 maio 2015
Ordem das Coisas
Mas aprendemos que a ordem dos fatores não afeta os produtos. Será? Quatro pesquisadores resolveram verificar se a ordem é relevante na vida. Semanalmente o National Bureau of Economic Research (NBER) distribui para mais de vinte mil pessoas a lista de novas pesquisas através de um e-mail. O NBER é extremamente conhecido na área de economia e geralmente os textos divulgados no sítio são posteriormente publicados nos periódicos mais relevantes das ciências sociais. Todos os textos são de altíssima qualidade.
O trabalho dos pesquisadores investigou o “primacy effect”, onde as pessoas selecionam os primeiros itens de uma lista. Existem diversas razões para que isto ocorra, como o cansaço, o efeito na memória ou a força da primeira impressão. O primacy effect foi investigado para verificar se a ordem dos artigos listados no e-mail do NBER teria algum efeito sobre o interesse dos leitores.
Apesar de serem trabalhos técnicos e não existir nenhuma preferência na ordem do e-mail encaminhado, a pesquisa conseguiu comprovar o primacy effect. Assim, os textos que estavam no topo da lista apresentaram maior número de downloads e citações em trabalhos futuros. E este efeito foi considerado significativo.
FEENBERG, Daniel; GANGULI, Ina; GAULE, Patrick; GRUBER, Jonathan. It´s Good to be First: Order bias in Reading and citing NBER Working Paper, maio de 2015.
27 maio 2015
Vladmir Vapnik, o gênio da matemática e computação
Computers and automated machines such as robots have reached an enormous level of sophistication, but they can't actually think—yet. But they can remember, recognize patterns, and make inferences and deductions, all in ways that can be used in real-world industrial, commercial, and scientific applications. Such phenomena are the basis of artificial intelligence, the branch of computer science that encompasses the increasingly important field of machine learning: the creation and development of algorithms that allow machines to learn from new data and change their behavior accordingly. Perhaps the most influential and innovative researcher in machine learning is Vladimir Vapnik, whose career and accomplishments practically define the discipline's current state of the art.
Machine learning and artificial intelligence draw heavily from mathematics and statistics. Vapnik obtained his M.S. in the former field in 1958 from Uzbek State University and his doctorate in the latter field in 1964 from Moscow's Institute of Control Sciences, where he would later become head of the computer science department. It was here that he began the work that ultimately led to his development, in collaboration with Alexey Chervonenkis, of Vapnik-Chervonenkis (VC) theory, which uses statistical and mathematical methods to explain the learning process, establishing the foundations of contemporary machine learning theory.
The seminal importance of Vapnik;s work did not begin to be fully recognized until he had the opportunity to leave Soviet Russia for an extended visit to the United States in 1989. He emigrated permanently to the U.S. in 1991 to take up a position at AT&T Bell Labs. At Bell, Vapnik continued to develop and build upon the ideas and implications of VC theory, finally inventing the concept of the support vector machine (SVM), a model and algorithm that allows a computer to identify and predict patterns and classify input into particular categories.
Its mathematically complex underpinnings may make machine learning appear to be mostly an abstract theoretical exercise, but Vapnik's research and particularly his SVM concept have led to a staggering variety of practical everyday applications. Machine learning algorithms are at the heart of fraud detection in electronic credit card transactions, computer security, speech and handwriting recognition, computerized medical diagnosis, DNA analysis, cataloging and data mining, and a host of other critical functions involving the identification and recognition of patterns and the classification of different types of information. In these and many other functions, Vapnik's work has moved directly from abstract theory to practical use.
2012 Ben Franklin Award Winner for Computer and Cognitive Science For more information, please visit https://www.fi.edu/laureates/vladimir-vapnik
One of the most fascinating aspects of artificial intelligence and machine learning research is that it can provide insight into one of science's most profound mysteries: the workings of our own brains and consciousness. Although these questions are not the focus of Vapnik's research, his models of the processes at work during learning and pattern recognition provide another perspective guiding the efforts of scientists studying how the human brain organizes and performs those functions.
Currently Vapnik is a professor of computer science and statistics at Royal Holloway, University of London, and holds a professorship in computer science at Columbia University in New York. Also working on staff at NEC Labs in Princeton, New Jersey, he continues to forge new paths in advanced mathematics, statistics, and their interactions and interconnections within computer science. His honors and awards include election to the National Academy of Engineering in 1996, the 2008 Paris Kanellakis Award, and the Alexander Humboldt Research Award for Lifetime Achievement.
While the question of whether computers will ever be able to think as human beings do may never be truly answered, there is no question that Vladimir Vapnik has invented ways to make them "think" better for all the myriad ways in which we humans use them every day.
Fonte: aqui
Assinar:
Postagens (Atom)
Postagens
