<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:wikidot="http://www.wikidot.com/rss-namespace">

	<channel>
		<title>Recitation 8 - gradient direction</title>
		<link>http://rl-tau-2019.wikidot.com/forum/t-12187353/recitation-8-gradient-direction</link>
		<description>Posts in the discussion thread &quot;Recitation 8 - gradient direction&quot;</description>
				<copyright></copyright>
		<lastBuildDate>Sun, 19 Apr 2026 23:37:45 +0000</lastBuildDate>
		
					<item>
				<guid>http://rl-tau-2019.wikidot.com/forum/t-12187353#post-4298600</guid>
				<title>(no title)</title>
				<link>http://rl-tau-2019.wikidot.com/forum/t-12187353/recitation-8-gradient-direction#post-4298600</link>
				<description></description>
				<pubDate>Fri, 05 Jul 2019 09:02:42 +0000</pubDate>
				<wikidot:authorName>Just asking</wikidot:authorName>								<content:encoded>
					<![CDATA[
						 <p>I take it back. I see that the two minus signs are canceled since you take f-f' and not f'-f.<br /> Thanks</p> 
				 	]]>
				</content:encoded>							</item>
					<item>
				<guid>http://rl-tau-2019.wikidot.com/forum/t-12187353#post-4298556</guid>
				<title>Recitation 8 - gradient direction</title>
				<link>http://rl-tau-2019.wikidot.com/forum/t-12187353/recitation-8-gradient-direction#post-4298556</link>
				<description></description>
				<pubDate>Fri, 05 Jul 2019 07:49:37 +0000</pubDate>
				<wikidot:authorName>just asking</wikidot:authorName>								<content:encoded>
					<![CDATA[
						 <p>Correct me if I am wrong but should the updates the weights of the Q function in the direction of the negative gradient.<br /> it seems that we add the gradient to the weights instead of subtracting it.</p> <p>Thanks</p> 
				 	]]>
				</content:encoded>							</item>
				</channel>
</rss>