tag:blogger.com,1999:blog-2264690864064246897.post4086597069575310015..comments2008-04-09T03:11:26.060-07:00Comments on Linux Tips & Tutorials: Speed up Gnome in Ubuntu: Tip2Heraldhttp://www.blogger.com/profile/13055364153724100269noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-2264690864064246897.post-30366136389129231052008-04-09T03:11:00.000-07:002008-04-09T03:11:00.000-07:00A quick aside: why use a Markov chain and not a Ba...A quick aside: why use a Markov chain and not a Bayesian network. Markov chain assume that the previous states are irrelevant and only use the present state to predicting the future state, which indeed is true at startup. However if a do a multitude of tasks that use overlapping sets of applications, then I will preload many unneeded libraries. Also, at first thought, the volume of data needed to consider each used set of libraries is O(2^n) whereas for a Bayesian network is O(n^2), n being the number of libraries.Heraldhttps://www.blogger.com/profile/13055364153724100269noreply@blogger.com