1 Ocak 2018 Dergi makalesi Açık Erişim

Gokcesu, Kaan; Kozat, Suleyman S.

MARC21 XML

<?xml version='1.0' encoding='UTF-8'?>
<record xmlns="http://www.loc.gov/MARC21/slim">
  <leader>00000nam##2200000uu#4500</leader>
  <datafield tag="245" ind1=" " ind2=" ">
    <subfield code="a">Online Density Estimation of Nonstationary Sources Using Exponential Family of Distributions</subfield>
  </datafield>
  <datafield tag="909" ind1="C" ind2="4">
    <subfield code="p">IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS</subfield>
    <subfield code="v">29</subfield>
    <subfield code="n">9</subfield>
    <subfield code="c">4473-4478</subfield>
  </datafield>
  <controlfield tag="001">30685</controlfield>
  <datafield tag="980" ind1=" " ind2=" ">
    <subfield code="a">user-tubitak-destekli-proje-yayinlari</subfield>
  </datafield>
  <datafield tag="520" ind1=" " ind2=" ">
    <subfield code="a">We investigate online probability density estimation (or learning) of nonstationary (and memoryless) sources using exponential family of distributions. To this end, we introduce a truly sequential algorithm that achieves Hannan-consistent log-loss regret performance against true probability distribution without requiring any information about the observation sequence (e.g., the time horizon T and the drift of the underlying distribution C) to optimize its parameters. Our results are guaranteed to hold in an individual sequence manner. Our log-loss performance with respect to the true probability density has regret bounds of O((CT)(1/2)), where C is the total change (drift) in the natural parameters of the underlying distribution. To achieve this, we design a variety of probability density estimators with exponentially quantized learning rates and merge them with a mixture-of-experts notion. Hence, we achieve this square-root regret with computational complexity only logarithmic in the time horizon. Thus, our algorithm can be efficiently used in big data applications. Apart from the regret bounds, through synthetic and real-life experiments, we demonstrate substantial performance gains with respect to the state-of-the-art probability density estimation algorithms in the literature.</subfield>
  </datafield>
  <datafield tag="650" ind1="1" ind2="7">
    <subfield code="2">opendefinition.org</subfield>
    <subfield code="a">cc-by</subfield>
  </datafield>
  <datafield tag="700" ind1=" " ind2=" ">
    <subfield code="u">Bilkent Univ, Dept Elect &amp; Elect Engn, TR-06800 Ankara, Turkey</subfield>
    <subfield code="a">Kozat, Suleyman S.</subfield>
  </datafield>
  <datafield tag="980" ind1=" " ind2=" ">
    <subfield code="b">article</subfield>
    <subfield code="a">publication</subfield>
  </datafield>
  <datafield tag="542" ind1=" " ind2=" ">
    <subfield code="l">open</subfield>
  </datafield>
  <datafield tag="100" ind1=" " ind2=" ">
    <subfield code="u">Bilkent Univ, Dept Elect &amp; Elect Engn, TR-06800 Ankara, Turkey</subfield>
    <subfield code="a">Gokcesu, Kaan</subfield>
  </datafield>
  <datafield tag="260" ind1=" " ind2=" ">
    <subfield code="c">2018-01-01</subfield>
  </datafield>
  <controlfield tag="005">20210315181757.0</controlfield>
  <datafield tag="909" ind1="C" ind2="O">
    <subfield code="o">oai:zenodo.org:30685</subfield>
    <subfield code="p">user-tubitak-destekli-proje-yayinlari</subfield>
  </datafield>
  <datafield tag="856" ind1="4" ind2=" ">
    <subfield code="z">md5:4adfbdf082d0064a5a6eeff8585dea3c</subfield>
    <subfield code="s">202</subfield>
    <subfield code="u">https://aperta.ulakbim.gov.trrecord/30685/files/bib-f827def4-2cc7-4c5c-8f8d-70d295027372.txt</subfield>
  </datafield>
  <datafield tag="540" ind1=" " ind2=" ">
    <subfield code="u">http://www.opendefinition.org/licenses/cc-by</subfield>
    <subfield code="a">Creative Commons Attribution</subfield>
  </datafield>
  <datafield tag="024" ind1=" " ind2=" ">
    <subfield code="a">10.1109/TNNLS.2017.2740003</subfield>
    <subfield code="2">doi</subfield>
  </datafield>
</record>