<?xml version="1.0" encoding="UTF-8"?>
<collection xmlns="http://www.loc.gov/MARC21/slim">
	<record>
		<datafield tag="980" ind1=" " ind2=" ">
			<subfield code="a">REPORT</subfield>
		</datafield>
		<datafield tag="970" ind1=" " ind2=" ">
			<subfield code="a">dimitrak-bengio_04-72/IDIAP</subfield>
		</datafield>
		<datafield tag="245" ind1=" " ind2=" ">
			<subfield code="a">Estimates of Parameter Distributions for Optimal Action Selection</subfield>
		</datafield>
		<datafield tag="700" ind1=" " ind2=" ">
			<subfield code="a">Dimitrakakis, Christos</subfield>
		</datafield>
		<datafield tag="700" ind1=" " ind2=" ">
			<subfield code="a">Bengio, Samy</subfield>
		</datafield>
		<datafield tag="856" ind1="4" ind2="0">
			<subfield code="i">EXTERNAL</subfield>
			<subfield code="u">http://publications.idiap.ch/attachments/reports/2004/rr-04-72.pdf</subfield>
			<subfield code="x">PUBLIC</subfield>
		</datafield>
		<datafield tag="088" ind1=" " ind2=" ">
			<subfield code="a">Idiap-RR-72-2004</subfield>
		</datafield>
		<datafield tag="260" ind1=" " ind2=" ">
			<subfield code="c">2004</subfield>
			<subfield code="b">IDIAP</subfield>
		</datafield>
		<datafield tag="520" ind1=" " ind2=" ">
			<subfield code="a">We present a general method for maintaining estimates of the distribution of parameters in arbitrary models. This is then applied to the estimation of probability distribution over actions in value-based reinforcement learning. While this approach is similar to other techniques that maintain a confidence measure for action-values, it nevertheless offers a new insight into current techniques and reveals potential avenues of further research.</subfield>
		</datafield>
	</record>
</collection>