<?xml version="1.0" encoding="UTF-8"?><!-- generator="WordPress/2.8.1" -->
<rss version="0.92">
<channel>
	<title>Trahald.com</title>
	<link>http://www.trahald.com</link>
	<description>Just another WordPress weblog</description>
	<lastBuildDate>Fri, 24 Jul 2009 11:03:36 +0000</lastBuildDate>
	<docs>http://backend.userland.com/rss092</docs>
	<language>en</language>
	
	<item>
		<title>Bug in model oddly causes improvement</title>
		<description>I created a pseudo-SVD++(3) model (mf_time.cpp in my framework) - one without the &#124;N(u)&#124;^(-.5)*Yi feedback term, and also without the alpha_u*dev_u_hat term (which I found to make the RMSE worse). However the one I initially wrote had a bug in the training and use of the alpha_u_k*dev_u_hat term (the term ...</description>
		<link>http://www.trahald.com/bug-in-model-oddly-causes-improvement/</link>
			</item>
	<item>
		<title>Kadri Framework</title>
		<description>The Kadri Framework is meant for the analysis of the Netflix data set. It is based on Icefox's Netflix Recommender Framework. It uses Icefox's framework to build the core data files, and expands on it to allow pre-processing, the use of dates, and blending.

Requirements:

	Qt
	LAPACK
	64-bit OS (or 32-bit OS with increased ...</description>
		<link>http://www.trahald.com/kadri-framework/</link>
			</item>
	<item>
		<title>kNN Source Sode</title>
		<description>Source code implementing the k-NN K Nearest Neighbors algorithm can be found at this post.

It implements Icefox's framework, details of which are in this thread. </description>
		<link>http://www.trahald.com/knn-source-sode/</link>
			</item>
</channel>
</rss>
