since i just did a large project project on image compression techniques, i'm really curious to know what they did to get the 28:1 odd compression ratios. based on my (admittedly limited) playing around with jpeg 2000, this seems like it would be an ideal application (in fact, i'm fairly certain they could squeeze even more data into that space at equivalent quality). in my experiments, jpeg 2000 beat the snot out of everything else (using the lovely jasper toolkit)... i don't know of too many other formats which can achieve 100:1 compression ratios in natural images and still have the results be recognizable. either way, very cool for an image processing geek like myself.
i haven't played with 8.0 yet, but using 7.1 i managed to get substantial (>20% overall) speedups. of course, this was with ipo turned on for almost everything which generated several megabytes worth of files per source file. i'm looking at playing with their fortran compiler, partially because it's a good excuse for me to play with fortran and partly because i'm fairly certain that it will be able squeeze some extra speed out of key algorithms. that said, even if the executables' speeds weren't substantially different, icc has some other nice features, built in openmp stuff, etc. and, of course, it's always good to have a second compiler's opinions on things.
you've either failed to communicate or you've demonstrated that neither you nor your it people have a clue.
sql = structured query language, i.e. a capability that rdbms may implement. access = m$ attempt at a rdbms 'lite' which implements part of sql sql is not synonymous with dbms, EVER.
i feel better now. that said, i'd stay the hell away from access if your goal is to teach a *database* class. if students want to learn access, send them to a course about office. if they want to learn about databases, skip the fluffy nonesense and do things properly, i.e. in the environment in which they'll be working. this means a real rdbms (such as oracle, mysql when >4.1 is released, postgress, etc.) anything less is like proposing to train people to drive 18 wheelers using ford pickup trucks.
fwiw, our it and mis (academic) departments both use oracle heavily and the cs department uses oracle exclusively in its database classes. as far as i'm concerned, while access has its place, that place is far far far away from anybody whose goal is to actually learn the principles of databases. if you want a well trained monkey who can do basic database stuff, send them to any ms office class. if you want someone who actually understands what's going on and has a set of skills (specifically, the "core" ddl and dml parts of sql that implementations agree on) then you teach using a real database.
ever had to sit through a 2h40 long lecture? the first 45 minutes aren't too bad, the next 45 are bearable, the remaining 70 are torturous (particularly when the lectures run 7:20-10:00pm.) most profs are kind enough to give a break in the middle, but i've had classes with a few who just forget. if you think the length of the lecture is rough on you, it's doubly so for the prof.
secondly, 6 hours of material is approximately two weeks worth of lecture. think about how much missing even one lecture in the wrong class can screw up your understanding of the subject, regardless of how many notes and lecture outlines you get.
I would appreciate it if you would devote more diligence to understanding what I wrote before you try to evaluate its rightness or wrongness. all due diligence was devoted. did i say you were wrong? no. i said you were naive and the solution you proposed is incredibly impractical because it makes assumptions about human behaviour which are patently false. you are correct in that not having sex means that you will most likely not contract an std. except... wait... herpes can be contracted orally (ever wonder what a cold sore is?) and hiv isn't only spread via sexual contact. but, to be fair, this could significantly reduce the available infection vectors. whether it will or not is an entirely different question. if you think you'll cure hiv/aids in africa through abstinence alone, you are sadly mistaken. it won't work in the united states either, for the same reason. people have sex, they always will, even those leading the "no sex until marriage" parade. promiscuity is a fact of biology and, in particular, male sexuality. that's the whole point of the male reproductive strategy: spread your seed wide and far. darwin in action.
I didn't say anything about the government (or anybody) imposing prohibition on any action. I'm talking about individuals choosing abstinence, which is neither illogical nor impractical. for individuals, yes, this is neither impractical nor illogical. but for abstience to be a viable option, everyone has to be abstaining.
True, people have to want to live that lifestyle. It may appear impractical on a large scale today only because the entertainment media and the public education system are working to promote risky sexual behaviors and dangerous attitudes toward sex and relationships. never assume a motive any more sinister than profit margins.
All this effort to fight AIDS medically is like straining gnats while swallowing a camel. It's ignoring the elephant in the room. AIDS is a social problem, not a medical problem. It is thought to have started from beastiality, and it has propagated through acts of sodomy and fornication involving multiple sexual partners. If everybody would pair up with one sexual partner for life, the problem would be solved. We should be doing everything we can to promote this kind of relationship, which is uniquely both healthful and ultimately fulfilling. to review: how many africans have hiv? even if you could miraculously stop the spread of hiv, what about the people who already have it? contrary to popular belief, sex isn't the only way to contract hiv (ever stop to consider why you have to sign a release saying that if a medical worker is exposed to your blood that you agree to be tested for hiv?) secondly, the claim that hiv is the result of bestiality is unsubstantiated at best.
Oh, but we don't want to be "prudes." Fine, have it your way. Long live AIDS. what a marvel of compassionate conservativism. </sarcasm>
and a dose of common sense would tell you that this is the most phenominally impractical solution to a problem since prohibition was suggested in the 1910's. of course, i may be a bit biased, what with letting some form of logic and realism enter into my thought processes...
dr. duesburg has impressive credentials, but he's making an assumption that correlation implies a cause and effect relationship when this is not true. if you can show results in peer reviewed journals like nature, science, etc. that support his conclusions, then i'd be much more inclined to believe you; unfortunately, right now, he's one voice saying "no" in a choir singing "yes".
first off, you're incredibly naive if you believe that george mason is the only university that does this (if you don't believe me, check here for mit's take). every university, (but particularly places like george mason which is a young school with 28,000 odd students between the arlington, fairfax, and manassas campuses) needs sources of revenue. and guess what, that's what intellectual property generates: revenue. it's not like the money is going to some nameless, faceless corporation. in this case, it's going towards defraying the costs of running a huge university which, in turn, translates to more reasonable tuition for students.
secondly, you know nothing about what terms the patent will be licensed under, so to immediately start accusing george mason of being populated by souless, flesh eating, corporate zombies out to rule the world is ridiculous.
grow up. not all patents are bad and not all patent holders are evil.
it's been many-a-moon since i took my foundations of math class, and i made a point of forgetting this particular section as soon as i possibly could before it hurt my head. of course, i've been using a computer for so long that i'm half convinced that the reals are a hoax invented by physicists to make their sums easier;-)
1 is the canonical representation of 0.9999... in the decimal system. incidentally, it is also the canonical representation of 1.00000...1. just to throw a spanner in the works, what's the largest real number smaller than 1? most folks want to say 0.99999... any mathheads out there who've done calculus or analysis more recently than i want to take a crack at it? (fwiw, i think the supremum of the set of all numbers less than 1 is 1, but the set itself doesn't contain a maximum)
totally off topic, but nonethelessL obligatory simpsons quote: "You know me Marge, I like my beer cold, my TV hard, and my homosexuals, flammmmming..." -- Homer Simpson
odd.... i know that m$ was pushing office interoperability with visio, kinda positioning it in the same kind of class as publisher is in --- a second string player on the team. anyways, isn't this what virtual pc is for?
I don't know if MS has ported this to OS X yet, but it sounds like this is exactly what you want. Yes, it's an M$ product; however, it's reasonably good at what it does (sorry folks, Dia has a *long* way to go to catch up) and can be used for a lot more than "just" office layout (the pro edition has office layout stuff, ERD and UML as well, if I remember correctly).
is essentially what this boils down to. i wish there was an easy answer, but at the end of the day, you're the guy who's going to have to sleep with how you've chosen to resolve this on your conscience. my personal $1/(49+1) is that yes, you do have an obligation to say, "hey, there may be a better way -- look at this possibility", but that's where it ends: there is no cure for willful stupidity. if management chooses an inferior option even after having other choices presented to them, that's management's problem (and maybe a clue to find somewhere else to work).
honestly, if you do end up buying a text book, i'd buy a college text book, because with the exception of the text books i used in my ap classes (which were all college text books), my high school text books universally sucked hairy goat testicles.
better yet, if there's a college nearby, why not see about taking a "101" class or two as an extended studies student... many employers have plans set up for continuing education, so it may even be cheap for you;-)
if this is a major sticking point for you, there's no reason to use wizard generated code. i know that a lot of people find it useful, but for me personally, i would rather write this stuff by hand. of course, i've never built anything with more than maybe four dialogs...
17% of what? measuring qualitative data with quantitative measures makes very little sense. think about it: can a pizza be "17% more cheesy" than another pizza?
</rant>
this is probably why i'm not a social scientist;-)
How many programming jobs require a solid understanding of mathematics? Not many, it seems to me. Instead, programming requires a solid understanding of how to be logical in solving a problem you have never seen before.
I appologize right now for any toes I step on:
We'll agree to disagree on this. My (admittedly limited) experience working in industry proved to me that there is a big difference between a well trained monkey and someone I would want to hire to write code. I'll admit right up front that I'm a bit of a snob, but I really don't believe that you can really succeed in computer science without some serious mathematics. Unfortunately, most people who aren't mathematics majors make it through their degrees without ever really touching on what mathematics really is most people think it's all about solving equations and taking derivatives and such nonesense. It's not. Mathematics is primarily concerned with asking, "I accept these axioms to be true. What must be true as a consequence?", a skill crucial in pretty much every field. I've had to deal with altogether too many people who think something is true without being able to justify their beliefs. What's worse are the sheep who blindly accept what is said on faith alone. Do you want to deal with these people as team members on a project? Do you want to deal with these people as project managers?
Now, speaking for my research experience, I can guarantee you that there's no way I could do what I am doing right now without some hard core math. It's hard to do robotics without a pretty solid background in analytic geometry, and computer vision is pretty much impossible without some serious calculus and probability.
... that this virus should be released so close to Strom Thurmond's passing and the US Supreme Court's decision in Lawrence V. Texas... I think I smell a vast, sweeping Conservative conspiracy;-)
both robust unit testing and formal methods have their place --- formal methods for proving design and unit testing for proving implementation. however, it seems to me that the question of why formal methods aren't more widespread in the us boils down to a question of perceived cost v. perceived benefit. let's say it takes "n-nines" of reliability to keep enough of your customers satisfied enough to keep coming back to you. let's say you've got two designs for software which accomplishes a certain task, and both models (if implemented correctly) would result in a product that the customer would be sufficiently happy with. which model do you implement? odds are mostly likely that you'll go with the cheapest one, because that's what maximizes profits. (forget everything i just said if you are an academic or coding for the love of it). to borrow a line from an engineering prof: "perfection is the enemy of good enough, and it's good enough that sells"
these are things that shouldn't be too hard to implement (eg: using the matrix formulation for fully connected backpropagation networks rather than the graph model) and yet have large toy value:
1) acquire a cheap webcam (i recommend the not-so-cheap route of firewire + an orange micro ibot) and use it to do something neat (this is what i do for a living;-)) --- intel's opencv library (google for it) has all sorts of cool demos. possible applications: motion detection, object detection and recognition, maybe even face recognition?
2) playing with neural networks (personally, i recommend implementing your own neural network package and then using it to solve some kind of problem --- maybe use it together with #1 to recognize things?)
3) robots (www.kipr.org has a boatload of info on this, as well as how to get starter kits)
4) ???
if you'd like help with #1 or #2, give me a yell and i'll do what i can to help (i have a linux toolkit that makes talking to firewire cameras a reasonably pleasant task)
i was contemplating writing something that made ada look forgiving;-) or at least was enough of a bitch that students took the whole "do this in pencil and paper first and save yourself a lot of aggravation" thing seriously... but to do that, i suppose i'd have to make them do all their assignments in fortran on punch cards, and have them drive across town to submit a batch job then wait a day or two for the output...
Slashdot Mirror
since i just did a large project project on image compression techniques, i'm really curious to know what they did to get the 28:1 odd compression ratios. based on my (admittedly limited) playing around with jpeg 2000, this seems like it would be an ideal application (in fact, i'm fairly certain they could squeeze even more data into that space at equivalent quality). in my experiments, jpeg 2000 beat the snot out of everything else (using the lovely jasper toolkit)... i don't know of too many other formats which can achieve 100:1 compression ratios in natural images and still have the results be recognizable. either way, very cool for an image processing geek like myself.
i haven't played with 8.0 yet, but using 7.1 i managed to get substantial (>20% overall) speedups. of course, this was with ipo turned on for almost everything which generated several megabytes worth of files per source file. i'm looking at playing with their fortran compiler, partially because it's a good excuse for me to play with fortran and partly because i'm fairly certain that it will be able squeeze some extra speed out of key algorithms. that said, even if the executables' speeds weren't substantially different, icc has some other nice features, built in openmp stuff, etc. and, of course, it's always good to have a second compiler's opinions on things.
you've either failed to communicate or you've demonstrated that neither you nor your it people have a clue.
sql = structured query language, i.e. a capability that rdbms may implement.
access = m$ attempt at a rdbms 'lite' which implements part of sql
sql is not synonymous with dbms, EVER.
i feel better now. that said, i'd stay the hell away from access if your goal is to teach a *database* class. if students want to learn access, send them to a course about office. if they want to learn about databases, skip the fluffy nonesense and do things properly, i.e. in the environment in which they'll be working. this means a real rdbms (such as oracle, mysql when >4.1 is released, postgress, etc.) anything less is like proposing to train people to drive 18 wheelers using ford pickup trucks.
you're completely right.
fwiw, our it and mis (academic) departments both use oracle heavily and the cs department uses oracle exclusively in its database classes. as far as i'm concerned, while access has its place, that place is far far far away from anybody whose goal is to actually learn the principles of databases. if you want a well trained monkey who can do basic database stuff, send them to any ms office class. if you want someone who actually understands what's going on and has a set of skills (specifically, the "core" ddl and dml parts of sql that implementations agree on) then you teach using a real database.
ever had to sit through a 2h40 long lecture? the first 45 minutes aren't too bad, the next 45 are bearable, the remaining 70 are torturous (particularly when the lectures run 7:20-10:00pm.) most profs are kind enough to give a break in the middle, but i've had classes with a few who just forget. if you think the length of the lecture is rough on you, it's doubly so for the prof.
secondly, 6 hours of material is approximately two weeks worth of lecture. think about how much missing even one lecture in the wrong class can screw up your understanding of the subject, regardless of how many notes and lecture outlines you get.
not a good plan if you ask me.
I would appreciate it if you would devote more diligence to understanding what I wrote before you try to evaluate its rightness or wrongness. all due diligence was devoted. did i say you were wrong? no. i said you were naive and the solution you proposed is incredibly impractical because it makes assumptions about human behaviour which are patently false. you are correct in that not having sex means that you will most likely not contract an std. except... wait... herpes can be contracted orally (ever wonder what a cold sore is?) and hiv isn't only spread via sexual contact. but, to be fair, this could significantly reduce the available infection vectors. whether it will or not is an entirely different question. if you think you'll cure hiv/aids in africa through abstinence alone, you are sadly mistaken. it won't work in the united states either, for the same reason. people have sex, they always will, even those leading the "no sex until marriage" parade. promiscuity is a fact of biology and, in particular, male sexuality. that's the whole point of the male reproductive strategy: spread your seed wide and far. darwin in action.
I didn't say anything about the government (or anybody) imposing prohibition on any action. I'm talking about individuals choosing abstinence, which is neither illogical nor impractical. for individuals, yes, this is neither impractical nor illogical. but for abstience to be a viable option, everyone has to be abstaining.
True, people have to want to live that lifestyle. It may appear impractical on a large scale today only because the entertainment media and the public education system are working to promote risky sexual behaviors and dangerous attitudes toward sex and relationships. never assume a motive any more sinister than profit margins.
All this effort to fight AIDS medically is like straining gnats while swallowing a camel. It's ignoring the elephant in the room. AIDS is a social problem, not a medical problem. It is thought to have started from beastiality, and it has propagated through acts of sodomy and fornication involving multiple sexual partners. If everybody would pair up with one sexual partner for life, the problem would be solved. We should be doing everything we can to promote this kind of relationship, which is uniquely both healthful and ultimately fulfilling. to review: how many africans have hiv? even if you could miraculously stop the spread of hiv, what about the people who already have it? contrary to popular belief, sex isn't the only way to contract hiv (ever stop to consider why you have to sign a release saying that if a medical worker is exposed to your blood that you agree to be tested for hiv?) secondly, the claim that hiv is the result of bestiality is unsubstantiated at best.
Oh, but we don't want to be "prudes." Fine, have it your way. Long live AIDS. what a marvel of compassionate conservativism. </sarcasm>
which is clearly why so many africans are hiv positive, becuase they're all gay. grow up or at least find a better way of justifying your bigotry.
and a dose of common sense would tell you that this is the most phenominally impractical solution to a problem since prohibition was suggested in the 1910's. of course, i may be a bit biased, what with letting some form of logic and realism enter into my thought processes...
dr. duesburg has impressive credentials, but he's making an assumption that correlation implies a cause and effect relationship when this is not true. if you can show results in peer reviewed journals like nature, science, etc. that support his conclusions, then i'd be much more inclined to believe you; unfortunately, right now, he's one voice saying "no" in a choir singing "yes".
first off, you're incredibly naive if you believe that george mason is the only university that does this (if you don't believe me, check here for mit's take). every university, (but particularly places like george mason which is a young school with 28,000 odd students between the arlington, fairfax, and manassas campuses) needs sources of revenue. and guess what, that's what intellectual property generates: revenue. it's not like the money is going to some nameless, faceless corporation. in this case, it's going towards defraying the costs of running a huge university which, in turn, translates to more reasonable tuition for students.
secondly, you know nothing about what terms the patent will be licensed under, so to immediately start accusing george mason of being populated by souless, flesh eating, corporate zombies out to rule the world is ridiculous.
grow up. not all patents are bad and not all patent holders are evil.
if you've been using a computer since you were ten, you should have a firm belief in groups, rings, and fields, just of the finite variety ;-P
it's been many-a-moon since i took my foundations of math class, and i made a point of forgetting this particular section as soon as i possibly could before it hurt my head. of course, i've been using a computer for so long that i'm half convinced that the reals are a hoax invented by physicists to make their sums easier ;-)
1 is the canonical representation of 0.9999... in the decimal system. incidentally, it is also the canonical representation of 1.00000...1. just to throw a spanner in the works, what's the largest real number smaller than 1? most folks want to say 0.99999... any mathheads out there who've done calculus or analysis more recently than i want to take a crack at it? (fwiw, i think the supremum of the set of all numbers less than 1 is 1, but the set itself doesn't contain a maximum)
totally off topic, but nonethelessL obligatory simpsons quote: "You know me Marge, I like my beer cold, my TV hard, and my homosexuals, flammmmming..." -- Homer Simpson
odd.... i know that m$ was pushing office interoperability with visio, kinda positioning it in the same kind of class as publisher is in --- a second string player on the team. anyways, isn't this what virtual pc is for?
I don't know if MS has ported this to OS X yet, but it sounds like this is exactly what you want. Yes, it's an M$ product; however, it's reasonably good at what it does (sorry folks, Dia has a *long* way to go to catch up) and can be used for a lot more than "just" office layout (the pro edition has office layout stuff, ERD and UML as well, if I remember correctly).
Just my $0.02.
is essentially what this boils down to. i wish there was an easy answer, but at the end of the day, you're the guy who's going to have to sleep with how you've chosen to resolve this on your conscience. my personal $1/(49+1) is that yes, you do have an obligation to say, "hey, there may be a better way -- look at this possibility", but that's where it ends: there is no cure for willful stupidity. if management chooses an inferior option even after having other choices presented to them, that's management's problem (and maybe a clue to find somewhere else to work).
honestly, if you do end up buying a text book, i'd buy a college text book, because with the exception of the text books i used in my ap classes (which were all college text books), my high school text books universally sucked hairy goat testicles.
;-)
better yet, if there's a college nearby, why not see about taking a "101" class or two as an extended studies student... many employers have plans set up for continuing education, so it may even be cheap for you
just my $1/50
if this is a major sticking point for you, there's no reason to use wizard generated code. i know that a lot of people find it useful, but for me personally, i would rather write this stuff by hand. of course, i've never built anything with more than maybe four dialogs...
<rant>
17% of what? measuring qualitative data with quantitative measures makes very little sense. think about it: can a pizza be "17% more cheesy" than another pizza?
</rant>
this is probably why i'm not a social scientist ;-)
I appologize right now for any toes I step on:
We'll agree to disagree on this. My (admittedly limited) experience working in industry proved to me that there is a big difference between a well trained monkey and someone I would want to hire to write code. I'll admit right up front that I'm a bit of a snob, but I really don't believe that you can really succeed in computer science without some serious mathematics. Unfortunately, most people who aren't mathematics majors make it through their degrees without ever really touching on what mathematics really is most people think it's all about solving equations and taking derivatives and such nonesense. It's not. Mathematics is primarily concerned with asking, "I accept these axioms to be true. What must be true as a consequence?", a skill crucial in pretty much every field. I've had to deal with altogether too many people who think something is true without being able to justify their beliefs. What's worse are the sheep who blindly accept what is said on faith alone. Do you want to deal with these people as team members on a project? Do you want to deal with these people as project managers?
Now, speaking for my research experience, I can guarantee you that there's no way I could do what I am doing right now without some hard core math. It's hard to do robotics without a pretty solid background in analytic geometry, and computer vision is pretty much impossible without some serious calculus and probability.
... that this virus should be released so close to Strom Thurmond's passing and the US Supreme Court's decision in Lawrence V. Texas... I think I smell a vast, sweeping Conservative conspiracy ;-)
both robust unit testing and formal methods have their place --- formal methods for proving design and unit testing for proving implementation. however, it seems to me that the question of why formal methods aren't more widespread in the us boils down to a question of perceived cost v. perceived benefit. let's say it takes "n-nines" of reliability to keep enough of your customers satisfied enough to keep coming back to you. let's say you've got two designs for software which accomplishes a certain task, and both models (if implemented correctly) would result in a product that the customer would be sufficiently happy with. which model do you implement? odds are mostly likely that you'll go with the cheapest one, because that's what maximizes profits. (forget everything i just said if you are an academic or coding for the love of it). to borrow a line from an engineering prof: "perfection is the enemy of good enough, and it's good enough that sells"
these are things that shouldn't be too hard to implement (eg: using the matrix formulation for fully connected backpropagation networks rather than the graph model) and yet have large toy value:
;-)) --- intel's opencv library (google for it) has all sorts of cool demos. possible applications: motion detection, object detection and recognition, maybe even face recognition?
1) acquire a cheap webcam (i recommend the not-so-cheap route of firewire + an orange micro ibot) and use it to do something neat (this is what i do for a living
2) playing with neural networks (personally, i recommend implementing your own neural network package and then using it to solve some kind of problem --- maybe use it together with #1 to recognize things?)
3) robots (www.kipr.org has a boatload of info on this, as well as how to get starter kits)
4) ???
if you'd like help with #1 or #2, give me a yell and i'll do what i can to help (i have a linux toolkit that makes talking to firewire cameras a reasonably pleasant task)
i was contemplating writing something that made ada look forgiving ;-) or at least was enough of a bitch that students took the whole "do this in pencil and paper first and save yourself a lot of aggravation" thing seriously... but to do that, i suppose i'd have to make them do all their assignments in fortran on punch cards, and have them drive across town to submit a batch job then wait a day or two for the output...