28 ms·
Both. I prefer the former; the latter pays better at the moment, I'm sorry to say. That may just be a reflection of the fact that I'm not yet where I'd like t
by apathy 19y ago
Both. I prefer the former; the latter pays better at the moment, I'm sorry to say. That may just be a reflection of the fact that I'm not yet where I'd like to be WRT the former.
- aswanson 19y agoI prefer the former Don't we all...The latter is a solved problem with a million variants, the former is new ground. Would you recommend any papers/books in that area?
- apathy 19y agoIn my area the most recent terribly-interesting paper was by the deCode people re: genetics of gene expression and inferring network 'modules' from patterns of activity. There are more theoretically-interesting methods to deduce such graph-based causal networks (eg. Bayesian networks sampled via Metropolis-Hastings MCMC, or the paradoxically 'dumber' but more viable-in-the-cloud methods using MLEM, of which I am a proponent) but the deCode guys have the advantage of actually producing useful results right now. Overall (i.e. not restricted to my field of expertise) I would suggest that the biggest trove of ready-to-be-applied material on the Web is Andrew Moore's site at CMU: http://www.autonlab.org/tutorials/index.html http://www.autonlab.org/tutorials/index.html Possibly the worst-kept secret on the Web :-)