<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small">Prezados(as):</div><div class="gmail_default" style="font-family:verdana,sans-serif;font-size:small"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif"><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Gostaria de convidá-los a participar uma sequência de Seminários em: </div><div class="gmail_default" style="color:rgb(0,0,0)"><font color="#131313" style="font-family:serif"><font face="Lucida Grande"><font><b>Kernel-based learning methods: </b>Applications toward </font></font></font><font face="Lucida Grande"><font>Inference for Large Scale Text Classification</font></font>.</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)"><br></div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Horário: quartas-feiras às 13 as 14hs.</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Local: Sala 221 IMECC.</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)"><br></div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Interessados: RSVP a <a href="mailto:dias@ime.unicamp.br">dias@ime.unicamp.br</a></div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Espaço limitado a 15 participantes.</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Após o primeiro mês, é esperado de cada participante apresentar um seminário sobre assunto.</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)"><br></div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)">Dentro vários, cito:</div><div class="gmail_default" style="font-size:small;color:rgb(0,0,0)"><p align="LEFT" style="margin-bottom:0in"><font face="Lucida Grande"><font style="font-size:13pt"><font color="#131313">Parte do Abstract do livro: </font><font face="serif"><i>I</i></font><font face="serif"><i><b>nductive Inference for</b></i></font></font></font></p><p style="margin-bottom:0in"><font face="serif"><font style="font-size:13pt"><i><b>Large Scale Text Classification. </b></i>Catarina Silva e Bernadete Ribeiro</font></font></p><p style="margin-bottom:0in"><br></p><p align="LEFT" style="margin-bottom:0in"><font color="#131313"><font face="Lucida Grande"><font style="font-size:13pt">Text classification is becoming a crucial task to analysts in different areas. In the last few decades the production of textual documents in digital form has increased exponentially. Their applications range from web pages to scientific documents, including emails, news and books. Searching for a digital text in Google is now more than a reality, it is a commonplace. In the near future, with the advent of intelligent text classification methods, people will have even more access to a large variety of enhanced digital text services, viz. filtering, searching and filing.</font></font></font></p><p align="LEFT" style="margin-bottom:0in"><font color="#131313"><font face="Lucida Grande"><font style="font-size:13pt">Despite the widespread use of digital texts, handling them is inherently difficult - the large amount of data necessary to represent them and the subjectivity of classification complicate matters. Earlier research has addressed the extraction of information from relatively small collections of well-structured documents such as news wires and scientific publications.</font></font></font></p><p align="LEFT" style="margin-bottom:0in"><font face="Lucida Grande"><font style="font-size:13pt"><font color="#131313">Intelligent text classification methods, which rely heavily on machine learning algorithms, have the potential to supersede existing information retrieval techniques and provide superior facilities that will save time and money for users and companies, while providing a vital tool for dealing with the proliferation of digital texts they are faced with.</font></font></font></p></div><div class="gmail-yj6qo gmail-ajU" style="font-size:16px;margin:2px 0px 0px;color:rgb(0,0,0);font-family:arial,sans-serif"></div></div><br clear="all"><div><div class="gmail_default" style="font-family:verdana,sans-serif">Extraido de: <span style="font-family:helvetica"><b><i>Learning with Kernels: </i></b></span><b style="font-family:arial,sans-serif"><font face="verdana, sans-serif"><i>Support Vector Machines, Regularization, Optimization, and Beyond</i><div class="gmail_default" style="display:inline">.</div></font></b></div>
<p class="gmail-p2"><font face="verdana, sans-serif">Bernhard Scholkopf</font></p><div class="gmail_default" style="font-size:small;display:inline"><font face="verdana, sans-serif"> </font></div><font face="verdana, sans-serif">Alexander J. Smola</font><p></p><div class="gmail_default" style="font-size:small"><font face="verdana, sans-serif"><br></font></div>
<p class="gmail-p1"><font face="verdana, sans-serif"> One of the most fortunate situations a scientist can encounter is to </font></p><p class="gmail-p1"><font face="verdana, sans-serif">enter a field in</font></p><div class="gmail_default" style="display:inline"><font face="verdana, sans-serif"> </font></div><font face="verdana, sans-serif">its infancy. There is a large choice of topics to work on, and many of the issues<div class="gmail_default" style="font-size:small;display:inline"> </div>are conceptual rather than merely technical.<div class="gmail_default" style="font-size:small;display:inline">...</div> </font><p></p><p class="gmail-p1"><font face="verdana, sans-serif"> The scope of the field has now widened significantly, both</font></p><div class="gmail_default" style="font-size:small;display:inline"><font face="verdana, sans-serif"> </font></div><font face="verdana, sans-serif">in terms of new algorithms, such as kernel methods different to SVMs, and in<div class="gmail_default" style="font-size:small;display:inline"> </div>terms of a deeper theoretical understanding being gained. It has become clear<div class="gmail_default" style="font-size:small;display:inline"> </div>that kernel methods provide a framework for tackling some rather profound<div class="gmail_default" style="font-size:small;display:inline"> </div>issues in machine learning theory. At the same time, successful applications have<div class="gmail_default" style="font-size:small;display:inline"> </div>demonstrated that SVMs not only have a more solid foundation than artificial<div class="gmail_default" style="font-size:small;display:inline"> </div>neural networks, but are able to serve as a replacement for neural networks that<div class="gmail_default" style="font-size:small;display:inline"> </div>perform as well or better, in a wide variety of fields. Standard neural network an<div class="gmail_default" style="font-size:small;display:inline"> </div>pattern recognition textbooks have now started including chapters on SVMs <div class="gmail_default" style="font-size:small;display:inline">and </div><div class="gmail_default" style="font-size:small;display:inline"></div>kernel PCA</font><p></p><div class="gmail_default" style="display:inline">.</div></div><div><p></p><br></div>-- <br><div class="gmail_signature"><div dir="ltr"><div><div>Ronaldo Dias</div>Professor<div>Dept. of Statistics-IMECC, UNICAMP</div><div><a href="http://www.ime.unicamp.br/~dias" target="_blank">www.ime.unicamp.br/~dias</a></div><div><br></div></div></div></div>
</div>