No Robert Goddard yet? Rockets! And he got his start right around the same age as the kids. You could throw in Wernher von Braun for a counterpoint as well.
Actually, Bioconductor's original intent was to provide a platform for statisticians doing research on analyzing high-throughput experiments (as opposed BioPerl and friends which AFAIK originally arose to deal with sequence analysis and the myriad of tools that go with it). To date, this has principally meant microarray experiments and such, though there is also a lot of support code for manipulating things common to Bioinformatics (annotations, ontologies and so on). Its also used by a fair number of biologists these days to actually analyze their experiments which has led to the development of some simple GUI interfaces.
Have you ever considered that the article will be published in a journal and read by people without the benefit of two-way communication with the author(s)? If, for instance, reviewer confusion can be fixed with a simple explanation then why isn't that explanation in the article? Reviewer comments are just that: comments that will help you improve the article for resubmission, not some sort of scientific hazing ritual and anonymous communication may only serve to degrade the quality of publication.
If it really bothers you that much, have colleagues not involved in the project review papers before submission and do the same for them.
First off, the spam filters are actually classification algorithms, not filters---the name filter is incorrectly used almost exclusively by spam classification software--and worse yet they're really only referring to a specific classifier (the "Naive Bayes" algorithm) rather than to classifiers in general. "Bayesian" filters are things like Kalman Filters, Particle Filters and Hidden Markov Models which are used in any number of fields, but not really germane to the tasks you're asking about I think. Using "Bayesian Classification" in Google will probably yield more fruitful results.
It sounds like you want to extend the naive bayes classifier to more than two categories and, in the best case, learn new categories from the data. Both can be done and have been done with varying degrees of success. You might try here for some pointers to more information about how it is done (the algorithm itself has been around since the '60s---people only think its something new). Unfortunately for things like RSS and email you're going to run into two problems: you really want to do your classification on-line and your data are actually quite sparse and your prior is usually uninformative so its going to be hard to do the actual classification. But, who knows, its still an active topic of research.
I'll tell you why they got mechanical watches and didn't hack up a Linux watch:
1. Generally speaking digital watches are fugly. There's no Movado Digital Watch for a reason.
2. Commitment. This watch will ALWAYS run ~24h39m. You can give it to your grandkids. Your crap-ass programmable digital watch won't make it that far. Also, it can be made back into a 24h watch. There are no digital watch family heirlooms.
3. A mechanical watch is a thing of craftsmanship and beauty. A watch running Windows or Linux is cute for maybe 10 minutes then its a watch that does so many other things that they forgot the "tells time" part.
No, of course I don't, and Microsoft (or anyone else for that matter) isn't obligated/inclined to report this information. The business model is simply The Way It Is Done and you are correct---the stuff is a couple of years old now and I wouldn't be surprised to learn that their loss per unit is maybe 1/2 or 1/3 of what is was originally (IIRC the expected loss per unit at launch was something like $100-$150 so that would make the loss somewhere between $30-$75 now).
Unfortunately, Microsoft gets to buy in bulk and has a relatively close relationship with their manufacturer, something Joe Consumer doesn't get which compounds the problem of building a machine with the same price point.
Well, since Microsoft is taking a loss on the XBox hardware so you might not be able to accomplish that. Its pretty standard Gillette-style marketing practice (give the razor away and make a killing on the blades).
Of course the 'killing on the blades' part is proving to be more difficult than they anticipated, but so it goes.
If you're actually going to be out of touch, like in outer Mongolia or something, go get a manual 35mm SLR. They don't need batteries at all and the bodies are virtually indestructible. Film is easy to carry, high quality, cheap and you can have it converted into slides when you like.
If you're going to be "out of touch" in major metropolitan areas outside the US then don't sweat it---there are WAY more Internet Cafe's outside the U.S. and contrary to popular belief there is civilization in other countries... you know, power, hot and cold running water... that sort of thing.
First, it would help if you told us what they're going to make you take for the program.
Second, what is your background. Do you have a CS background? Biology? English?
As for the courses of those Large Datasets is probably the most useful. Image Classification? Depends. If by 'bioinformatics' your actually talking about Biomedical IS (things like Query By Doodle for retrieving X-Ray information and so on) then yeah, but working on sequence data there aren't any images to classify. Maybe for some of the labeling experiments, buts thats more a problem in segmentation not classification. I'd call it less useful.
I don't even know what you mean by Subsymbolic AI (search? more image analysis?). Depending on what you meant by that it could be useful. There's a good deal of crossover between text analysis and sequence analysis.
Parallel Programming? Not so much I'm thinking. Most algorithms you're interested in are trivially parallel so at most you need scatter-gather type parallel programs, unless by Bioinformatics you mean Computational Chemistry then you could run into trouble.
If you're background is Biology you'd be best served brushing up the Math skills (Statistics and Applied Mathematics, Mathematical Modeling or Biomedical Modeling if you can find it). Computer Science, Math (depending on your program) and Biology. Math? Biology.
English? Math, Biology and some intense programming experience.
Hopefully the GUI thing will improve over time--there's a project called ObveRsive (or something like that) thats gotten started within the last few months to put, from what I can tell, Stata/DataDesk types of things over R...There's been a great deal of interest in that from time to time.
There's also at least one person working on more interactive forms of documents (read: widgets and data in your paper controlling live graphing surfaces) which could be used to build GUIs as well. You can find info about some of this cutting edge stuff over at http://omegahat.org and I think http://obversive.sf.net
Until those arrive there's also Sweave (its in the tools package that ships), which while not really a GUI lets you combine LaTeX and R noweb style to generate some quite nice ps/pdf output (it deals with making sure the appropriate PDF or EPS files are around and whatnot).
Or RWeb which lets you build CGI based interfaces to R (complete with graphs), though performance isn't exactly stellar (R has a pretty heavy launch penalty relative to most computational tasks)
Well, that depends on how much work you want to do. It would be difficult for any package to do everything automatically since some of it is an aesthetic decision, but most good packages try. S-PLUS and R (GNU S) mentioned elsewhere in this post both try to do sort of the aesthetic thing---scaling for the 45 degree line and so on.
You can still make stuff look awful if you try though, you do have a complete programming language at your disposal if you try:-)
I use the R statistical package my self. Depending on the version you get it has either a Carbon or X11 interface (an Aqua interface is being built) and provides a number of different output formats including JPG, PNG, PS, PDF and (using an external package RSVGDevice) SVG.
http://www.r-project.org (has the official Carbon build, an X11 super package with something like 300 external packages is also occasionally built, check the R-SIG-Mac mailinglist for information about that)
Additionally Matlab (R for certain) is more likely to have pre-existing code written by Those Who Know What They Are Doing so you don't have to reinvent the wheel.
Matlab is probably more useful to you than Mathematica since you'll be working with simulations and/or data it sounds like. You might also want to check out R, which is designed as a statistical analysis environment and has a large number of packages---including machine learning and whatnot. Its also Open Source and GPL if you care about that sort of thing and runs on pretty much any platform you could potentially care about.
So, the dubbed version (which I saw last night) is just fine. They use unique voiceactors for each character and they convey the emotion well enough. Of course, if you speak Japanese.
Personally, I'm okay with dubbing--but then I also like to pick up the appopriate emotive cues, which is more difficult when people are speaking a language you don't understand. Of course, usually the dubbing is awful so having it in English doesn't do a bit of good. Oddly, I don't think the same thing about live action dubbing, but I think that's just because the loss of sync is WAY more annoying than not picking up tone of voice properly.
Right, surgical specialities aside (whom I appreciate greatly having been under-the-knife twice in the past 18 months with nothing untoward happening to me) this is more-or-less what European medicince does---emphasizes relationship building over pattern matching capability since the former is the 'hard bit,' whereas the latter is something that can be attacked via automation mechanisms. I'm not saying that computers can or even should replace humans in the task of diagnosis. However, hostility towards computer-assisted diagnostics seems silly (sort of like having a hammer but preferring to use the claw to hold the nails rather than pounding them...which you do with your bare hands)--you've got the tool and I'm willing to guess that MDs use search engines all the time (hell, there are even search engines for things like PubMed that research MDs probably couldn't live without at this point) for other things so why NOT for their jobs?
I would guess that isn't necessarily true--the culture of warez is probably just as much a draw as the technical challenge of a crack for those who partake (just look at courier groups. they move things. this is interesting? no, but it *does* have a suitable cultural environment). You'd have to figure out someway to make bug hunting an illicit rush of some sort. Not gonna happen.
So, they found that people sleeping longer are more likely to die. So what. That says *nothing* about the actual effect of sleeping---more plausible is that the more sick (sicker? whatever) you are the more you sleep since you're body is trying (perhaps vainly) to repair itself/conserve energy.
Seriously, this article practices once of the major fallacies of statistics.
Major fallacies of psychology, statistics had nothing to do with this study--- as should be obvious from the article. They showed that adding uses a different part of the brain than a video game that was more-likely-than-not designed to engage the run-from-predators-and-kill-food part of our brains. Science at its finest.
What I'd really like to know is how they identified the part of the brain responisible for 'controlling anti-social behaviour.' Saying that the frontal lobes are responsible for self-control is a bit like saying the brain is responsible for thinking (some believe it may be so).
RMS is somebody with a Vision that he wishes the world to share, whether or not it really wants to.
And this is different from being a politician how? As far as I can tell this is what politicians do all day-- attempt to foise there version of reality on my world whether I want it or not (usually the latter). RMS is different only in that a) he isn't very sneaky by goverment politician standards (backstabbing tactics only count as marginally sneaky, he obviously didn't play enough Supremacy as a kid) and b) his "target audience" tend towards apolitical free thinkers more than anything else so his preaching (for lack of a better word) just annoys people who ultimately want more or less the same thing, except with less insanity and foaming at the mouth (he obviously didn't go to enough RHPC as a kid either--- Don't Scare the Squares!), 'cause when you get right down to it there is only limited space on the lecture circuit and most of us will require real jobs to support families, video game habits, snowboarding expeditions, etc. (order to preference)
I would be more worried about bored sysadmins and I'm sure we ALL remember the original BOFH stories. I doubt the NSA is going to snoop my email to discover that I've been offered a fantastic deal on printer toner.
Slashdot Mirror
No Robert Goddard yet? Rockets! And he got his start right around the same age as the kids. You could throw in Wernher von Braun for a counterpoint as well.
Actually, Bioconductor's original intent was to provide a platform for statisticians doing research on analyzing high-throughput experiments (as opposed BioPerl and friends which AFAIK originally arose to deal with sequence analysis and the myriad of tools that go with it). To date, this has principally meant microarray experiments and such, though there is also a lot of support code for manipulating things common to Bioinformatics (annotations, ontologies and so on). Its also used by a fair number of biologists these days to actually analyze their experiments which has led to the development of some simple GUI interfaces.
Have you ever considered that the article will be published in a journal and read by people without the benefit of two-way communication with the author(s)? If, for instance, reviewer confusion can be fixed with a simple explanation then why isn't that explanation in the article? Reviewer comments are just that: comments that will help you improve the article for resubmission, not some sort of scientific hazing ritual and anonymous communication may only serve to degrade the quality of publication.
If it really bothers you that much, have colleagues not involved in the project review papers before submission and do the same for them.
First off, the spam filters are actually classification algorithms, not filters---the name filter is incorrectly used almost exclusively by spam classification software--and worse yet they're really only referring to a specific classifier (the "Naive Bayes" algorithm) rather than to classifiers in general. "Bayesian" filters are things like Kalman Filters, Particle Filters and Hidden Markov Models which are used in any number of fields, but not really germane to the tasks you're asking about I think. Using "Bayesian Classification" in Google will probably yield more fruitful results.
It sounds like you want to extend the naive bayes classifier to more than two categories and, in the best case, learn new categories from the data. Both can be done and have been done with varying degrees of success. You might try here for some pointers to more information about how it is done (the algorithm itself has been around since the '60s---people only think its something new). Unfortunately for things like RSS and email you're going to run into two problems: you really want to do your classification on-line and your data are actually quite sparse and your prior is usually uninformative so its going to be hard to do the actual classification. But, who knows, its still an active topic of research.
Those are quartz driven watches, not digital watches.
I'll tell you why they got mechanical watches and didn't hack up a Linux watch:
1. Generally speaking digital watches are fugly. There's no Movado Digital Watch for a reason.
2. Commitment. This watch will ALWAYS run ~24h39m. You can give it to your grandkids. Your crap-ass programmable digital watch won't make it that far. Also, it can be made back into a 24h watch. There are no digital watch family heirlooms.
3. A mechanical watch is a thing of craftsmanship and beauty. A watch running Windows or Linux is cute for maybe 10 minutes then its a watch that does so many other things that they forgot the "tells time" part.
No, of course I don't, and Microsoft (or anyone else for that matter) isn't obligated/inclined to report this information. The business model is simply The Way It Is Done and you are correct---the stuff is a couple of years old now and I wouldn't be surprised to learn that their loss per unit is maybe 1/2 or 1/3 of what is was originally (IIRC the expected loss per unit at launch was something like $100-$150 so that would make the loss somewhere between $30-$75 now).
Unfortunately, Microsoft gets to buy in bulk and has a relatively close relationship with their manufacturer, something Joe Consumer doesn't get which compounds the problem of building a machine with the same price point.
Well, since Microsoft is taking a loss on the XBox hardware so you might not be able to accomplish that. Its pretty standard Gillette-style marketing practice (give the razor away and make a killing on the blades).
Of course the 'killing on the blades' part is proving to be more difficult than they anticipated, but so it goes.
If you're actually going to be out of touch, like in outer Mongolia or something, go get a manual 35mm SLR. They don't need batteries at all and the bodies are virtually indestructible. Film is easy to carry, high quality, cheap and you can have it converted into slides when you like.
If you're going to be "out of touch" in major metropolitan areas outside the US then don't sweat it---there are WAY more Internet Cafe's outside the U.S. and contrary to popular belief there is civilization in other countries... you know, power, hot and cold running water... that sort of thing.
First, it would help if you told us what they're going to make you take for the program.
Second, what is your background. Do you have a CS background? Biology? English?
As for the courses of those Large Datasets is probably the most useful. Image Classification? Depends. If by 'bioinformatics' your actually talking about Biomedical IS (things like Query By Doodle for retrieving X-Ray information and so on) then yeah, but working on sequence data there aren't any images to classify. Maybe for some of the labeling experiments, buts thats more a problem in segmentation not classification. I'd call it less useful.
I don't even know what you mean by Subsymbolic AI (search? more image analysis?). Depending on what you meant by that it could be useful. There's a good deal of crossover between text analysis and sequence analysis.
Parallel Programming? Not so much I'm thinking. Most algorithms you're interested in are trivially parallel so at most you need scatter-gather type parallel programs, unless by Bioinformatics you mean Computational Chemistry then you could run into trouble.
If you're background is Biology you'd be best served brushing up the Math skills (Statistics and Applied Mathematics, Mathematical Modeling or Biomedical Modeling if you can find it). Computer Science, Math (depending on your program) and Biology. Math? Biology.
English? Math, Biology and some intense programming experience.
Hopefully the GUI thing will improve over time--there's a project called ObveRsive (or something like that) thats gotten started within the last few months to put, from what I can tell, Stata/DataDesk types of things over R...There's been a great deal of interest in that from time to time.
There's also at least one person working on more interactive forms of documents (read: widgets and data in your paper controlling live graphing surfaces) which could be used to build GUIs as well. You can find info about some of this cutting edge stuff over at http://omegahat.org and I think http://obversive.sf.net
Until those arrive there's also Sweave (its in the tools package that ships), which while not really a GUI lets you combine LaTeX and R noweb style to generate some quite nice ps/pdf output (it deals with making sure the appropriate PDF or EPS files are around and whatnot).
Or RWeb which lets you build CGI based interfaces to R (complete with graphs), though performance isn't exactly stellar (R has a pretty heavy launch penalty relative to most computational tasks)
Well, that depends on how much work you want to do. It would be difficult for any package to do everything automatically since some of it is an aesthetic decision, but most good packages try. S-PLUS and R (GNU S) mentioned elsewhere in this post both try to do sort of the aesthetic thing---scaling for the 45 degree line and so on.
:-)
You can still make stuff look awful if you try though, you do have a complete programming language at your disposal if you try
I use the R statistical package my self. Depending on the version you get it has either a Carbon or X11 interface (an Aqua interface is being built) and provides a number of different output formats including JPG, PNG, PS, PDF and (using an external package RSVGDevice) SVG.
http://www.r-project.org (has the official Carbon build, an X11 super package with something like 300 external packages is also occasionally built, check the R-SIG-Mac mailinglist for information about that)
Additionally Matlab (R for certain) is more likely to have pre-existing code written by Those Who Know What They Are Doing so you don't have to reinvent the wheel.
Matlab is probably more useful to you than Mathematica since you'll be working with simulations and/or data it sounds like. You might also want to check out R, which is designed as a statistical analysis environment and has a large number of packages---including machine learning and whatnot. Its also Open Source and GPL if you care about that sort of thing and runs on pretty much any platform you could potentially care about.
Also known as a coin...
So, the dubbed version (which I saw last night) is just fine. They use unique voiceactors for each character and they convey the emotion well enough. Of course, if you speak Japanese.
Personally, I'm okay with dubbing--but then I also like to pick up the appopriate emotive cues, which is more difficult when people are speaking a language you don't understand. Of course, usually the dubbing is awful so having it in English doesn't do a bit of good. Oddly, I don't think the same thing about live action dubbing, but I think that's just because the loss of sync is WAY more annoying than not picking up tone of voice properly.
Didn't MacDevcenter do something on that a few weeks ago?
Right, surgical specialities aside (whom I appreciate greatly having been under-the-knife twice in the past 18 months with nothing untoward happening to me) this is more-or-less what European medicince does---emphasizes relationship building over pattern matching capability since the former is the 'hard bit,' whereas the latter is something that can be attacked via automation mechanisms. I'm not saying that computers can or even should replace humans in the task of diagnosis. However, hostility towards computer-assisted diagnostics seems silly (sort of like having a hammer but preferring to use the claw to hold the nails rather than pounding them...which you do with your bare hands)--you've got the tool and I'm willing to guess that MDs use search engines all the time (hell, there are even search engines for things like PubMed that research MDs probably couldn't live without at this point) for other things so why NOT for their jobs?
I would guess that isn't necessarily true--the culture of warez is probably just as much a draw as the technical challenge of a crack for those who partake (just look at courier groups. they move things. this is interesting? no, but it *does* have a suitable cultural environment). You'd have to figure out someway to make bug hunting an illicit rush of some sort. Not gonna happen.
It's really for your own good, nobody wants to be beaten to death with their own laptop.
So, they found that people sleeping longer are more likely to die. So what. That says *nothing* about the actual effect of sleeping---more plausible is that the more sick (sicker? whatever) you are the more you sleep since you're body is trying (perhaps vainly) to repair itself/conserve energy.
correlation != causation
Seriously, this article practices once of the major fallacies of statistics.
Major fallacies of psychology, statistics had nothing to do with this study--- as should be obvious from the article. They showed that adding uses a different part of the brain than a video game that was more-likely-than-not designed to engage the run-from-predators-and-kill-food part of our brains. Science at its finest.
What I'd really like to know is how they identified the part of the brain responisible for 'controlling anti-social behaviour.' Saying that the frontal lobes are responsible for self-control is a bit like saying the brain is responsible for thinking (some believe it may be so).
RMS is somebody with a Vision that he wishes the world to share, whether or not it really wants to.
And this is different from being a politician how? As far as I can tell this is what politicians do all day-- attempt to foise there version of reality on my world whether I want it or not (usually the latter). RMS is different only in that a) he isn't very sneaky by goverment politician standards (backstabbing tactics only count as marginally sneaky, he obviously didn't play enough Supremacy as a kid) and b) his "target audience" tend towards apolitical free thinkers more than anything else so his preaching (for lack of a better word) just annoys people who ultimately want more or less the same thing, except with less insanity and foaming at the mouth (he obviously didn't go to enough RHPC as a kid either--- Don't Scare the Squares!), 'cause when you get right down to it there is only limited space on the lecture circuit and most of us will require real jobs to support families, video game habits, snowboarding expeditions, etc. (order to preference)
I would be more worried about bored sysadmins and I'm sure we ALL remember the original BOFH stories. I doubt the NSA is going to snoop my email to discover that I've been offered a fantastic deal on printer toner.