me, 2.0: meijer flower (6)

me, 2.0: jose nazario beauty and the street

meijer flower (6)

http://monkey.org/~jose/figs/IMG_1011.jpg

would you believe this is my digital elph S500 in macro mode, untouched? pretty awesome, isn't it? not quite SLR quality, but wow all the same.

one of the fallouts from my notacon talk was that people were talking about clustering methods. this is great, this is what i wanted people to do! they're learning about new things, and i couldn't be happier. i'm no expert on such things, i'm doing more like "grouping" or "lumping", but here's some pointers all the same:

open source clustering software
Multidocument summarization: An added value to clustering in interactive retrieval
Efficient Clustering of Very Large Document Collections

and finally, some stuff on Bayesian networks, the "other Bayesian technique" that few people every bother to learn about. damn you, paul graham, and your plan for spam. you've made algorithm wannabes all over the damned world abuse the Bayesian classification methods.
Learning Bayesian Network Structure from Massive Datasets: The "Sparse Candidate" Algorithm
A tutorial on learning with bayesian networks

i mention those last two because i've been looking at doing time-series analysis of the terms i group around from my RssClustering and seeing if there isn't a higher order to it all. so far i haven't implemented a useful network model builder, let alone any interesting results.

next Thursday, Apr 14, 2005 @ 07:02pm | previous Tuesday, Apr 12, 2005 @ 07:01pm | archives

Last modified: Wednesday, Apr 13, 2005 @ 07:02pm

Your Ad Here