Interesting. Usually, I would have expected Farkas to wiggle out with the help of $$ and serve little to no time, which is the case with almost every rich guy. But in this case, it seems that he is still serving time!
From Wikipedia:
Farkas is currently serving his sentence at the Federal Correctional Institution, Williamsburg, a facility in South Carolina that holds minimum and medium security inmates, and is scheduled for release in 2037
Religion? Eating meat? Utter nonsense. I know this is Slashdot and no one RTFA, but if you have a claim, you can quickly look it up using the Ctrl-F supercombo key.
India is following a global trend to phase out animal dissection, or to make it voluntary, despite opposition from scientists who say that the experience is impossible to replicate with models.
India’s ban comes after pressure from animal-rights groups such as the Indian arm of PETA (People for the Ethical Treatment of Animals), based in Mumbai, which used Bollywood celebrities in a high-profile campaign to bring college students on side
Why do you think R is not easy to implement? My company has been using SAS for a long time and we are finally making the change to R. As far as OP's requirements are concerned, I think R is way superior to SAS or SPSS because of its free, modular nature. It is clean, simple and suitable for a wide range of users. The commercial packages are filled with way too much business lingo garbage for me.
I personally think commercial support is overrated. I can install software on my own. I know how to browse through manuals and other information to find what I need. For a package like R, I almost always get any questions answered in at most few hours on online forums. So what exactly do I get from commercial support for my money?
But, if OP needs commercial support, there is an enterprise edition of R by Revolution Analytics located here: http://www.revolutionanalytics.com/products/revolution-enterprise.php. Might be worth looking into.
OP hasn't mentioned a lot except budget. Since you are on such a tight budget, I would highly recommend doing some theoretical analysis first. Do you have a serial code? How much parallelism exists in the code? You say the task is 'very parallel', but Amdahl's law (which is really common sense) will tell you that even for small amounts of serial sections of code, your speedup will be limited. You should also consider the amount of time the code actually runs. Achieving a speedup of 2 for a serial code that runs for one minute is near worthless.
After you estimate speedup, do some rough calculations on the basis of average cost of a processor and the the number of processors required. This should give you an estimate of the hardware cost required. Compare that with the cost of CPU cycles per dollar you get using a cloud service such as Amazon.
If Time Zones in Indiana are not broken, then I don't know what is. I mean, if the times could change if you commute from one place to another. Ridiculous.
Funny, as someone who is about to start working as a Statistician, all three courses add concrete value to my job - especially Machine Learning and Databases. Looking forward to the courses.
Yup, the physical distances cannot be ignored anymore. If fact, they try to place the servers as physically close to the corresponding stock exchange as possible in order to avoid speed-of-light delays!
This is purely my guesswork, but I think that the benefits would be marginal. An improvement of 1 fps might not be convincing for game devs to modify the kernel. However, if you can execute even a single trade before others do, and multiply it by the number of trades executed per second (which must be in millions), the profit margin would be non-trivial.
As mentioned in the summary, linux allows the firms to modify the OS kernel to serve their purpose. For example, the performance of trading algorithms would considerably degrade if context switching is allowed. So you can modify the kernel so as to dedicate certain cores for the main algorithms to which the OS can pass a very limited number of signals.
I know for a fact that at least one bank employs this in their high frequency trading group and probably all of them do.
To me, it seems like a simple Machine Learning problem. The number of common friends will obviously be the most influencing input parameter followed by school, work place, etc.
Unless someone makes me understand the complexities underneath, I am going to assume that anyone can solve this problem by taking one course in Machine Learning.
Exactly. It is also one of the reasons why some sites such as reddit start out being genuinely intellectual and as the user base grows, they progressively become just banal - which is what reddit is today.
It includes reviews by anyone who has never used the product. Example, this Zenith watch. I won't debate the hilarity of the reviews, but no doubt they are fake.
I think the "undecidability" factor comes into play in any machine learning problem. For example, which email is spam and which isn't? But significant strides have been made to solve that problem and I don't see why online review system is any different.
The current law stipulates that no more than 65,000 H1-B visas be issued each fiscal year. As of July 18, only 20,500 of these are filled. Of those, around 12,800 hold a Master's degrees. Are you suggesting that stopping the H1-B program is going to improve employment in US? No. What you'll be doing is reducing the quality of workers in American workplaces.
As regards to removing taxation, I am somewhat ambiguous. One could argue that since H1-B holders are not eligible to receive Social Security benefits, so they should not pay taxes However, a significant number of H1-B holders stay on to obtain a green card, making them eligible for the said benefits. Personally, I have no problem paying the Social Security taxes. The job that I am going to start in a month or so is fairly high-tech, the work is very intellectually stimulating and is simply not done in my home country. I am more than happy to help the currently unemployed Americans by paying my share of taxes. I view this as a price I pay for the better living conditions and greater job satisfaction. Mind you, that does not mean that either of them will be able to do my job. If any American is as skilled as I am, he won't have any problem getting a similar job position as mine.
The German movie Das Experiment is based on this experiment. Although they exaggerate quite a lot towards the end, first few days of the movie are real. Overall, an entertaining watch.
My search for a good version of tabbed putty still continues. Currently I'm using Superputty. Unfortunately, it is far from a finished product - the main problems being lack of proper keyboard shortcuts and non-regain of focus on maximize/minimize.
If someone knows a good tabbed putty version, my life would become a lot simpler.
Interestingly, software is affecting law industry big time. Gone are the days when lawyers needed to burn the midnight oil to research relevant cases. With a document managing software with proper tagging features, even you can become a lawyer!.
It was Dana Ballard. IIRC, the work that I referred to was done when he was at Rochester. He seems to have a link to the peanut butter and jelly video up on his page, but I am not able to download it.
I couldn't gather much from TFA apart from that they recognized the activities and the learning model behind it. I don't recall how much training data Dr. Ballard used, but the activity recognition in real time was very accurate.
My Machine Learning professor at University of Texas was hugely into Reinforcement Learning. He and one of his graduate student had developed this kind of system many years ago. Basically, only using a simple video camera, the computer could identify correctly all sub-activities involved in making a peanut butter sandwich; such as 'open jar', 'jelly on bread', 'jar closed', etc. We were shown the video during first class of the course and it was pretty cool.
The MSR guys are doing interesting stuff but this is nothing groundbreaking.
Slashdot Mirror
Farkas is currently serving his sentence at the Federal Correctional Institution, Williamsburg, a facility in South Carolina that holds minimum and medium security inmates, and is scheduled for release in 2037
Seriously, what's the difference between lobbying and bribery?
Religion? Eating meat? Utter nonsense. I know this is Slashdot and no one RTFA, but if you have a claim, you can quickly look it up using the Ctrl-F supercombo key.
India is following a global trend to phase out animal dissection, or to make it voluntary, despite opposition from scientists who say that the experience is impossible to replicate with models.
India’s ban comes after pressure from animal-rights groups such as the Indian arm of PETA (People for the Ethical Treatment of Animals), based in Mumbai, which used Bollywood celebrities in a high-profile campaign to bring college students on side
Why do you think R is not easy to implement? My company has been using SAS for a long time and we are finally making the change to R. As far as OP's requirements are concerned, I think R is way superior to SAS or SPSS because of its free, modular nature. It is clean, simple and suitable for a wide range of users. The commercial packages are filled with way too much business lingo garbage for me.
I personally think commercial support is overrated. I can install software on my own. I know how to browse through manuals and other information to find what I need. For a package like R, I almost always get any questions answered in at most few hours on online forums. So what exactly do I get from commercial support for my money? But, if OP needs commercial support, there is an enterprise edition of R by Revolution Analytics located here: http://www.revolutionanalytics.com/products/revolution-enterprise.php. Might be worth looking into.
Bottom line: R all the way.
OP hasn't mentioned a lot except budget. Since you are on such a tight budget, I would highly recommend doing some theoretical analysis first. Do you have a serial code? How much parallelism exists in the code? You say the task is 'very parallel', but Amdahl's law (which is really common sense) will tell you that even for small amounts of serial sections of code, your speedup will be limited. You should also consider the amount of time the code actually runs. Achieving a speedup of 2 for a serial code that runs for one minute is near worthless.
After you estimate speedup, do some rough calculations on the basis of average cost of a processor and the the number of processors required. This should give you an estimate of the hardware cost required. Compare that with the cost of CPU cycles per dollar you get using a cloud service such as Amazon.
Like what?
Kevin Mitnick was recently on Colbert Report to promote his book. Here is the link if anyone's interested.
If Time Zones in Indiana are not broken, then I don't know what is. I mean, if the times could change if you commute from one place to another. Ridiculous.
Funny, as someone who is about to start working as a Statistician, all three courses add concrete value to my job - especially Machine Learning and Databases. Looking forward to the courses.
Yup, the physical distances cannot be ignored anymore. If fact, they try to place the servers as physically close to the corresponding stock exchange as possible in order to avoid speed-of-light delays!
This is purely my guesswork, but I think that the benefits would be marginal. An improvement of 1 fps might not be convincing for game devs to modify the kernel. However, if you can execute even a single trade before others do, and multiply it by the number of trades executed per second (which must be in millions), the profit margin would be non-trivial.
As mentioned in the summary, linux allows the firms to modify the OS kernel to serve their purpose. For example, the performance of trading algorithms would considerably degrade if context switching is allowed. So you can modify the kernel so as to dedicate certain cores for the main algorithms to which the OS can pass a very limited number of signals.
I know for a fact that at least one bank employs this in their high frequency trading group and probably all of them do.
To me, it seems like a simple Machine Learning problem. The number of common friends will obviously be the most influencing input parameter followed by school, work place, etc.
Unless someone makes me understand the complexities underneath, I am going to assume that anyone can solve this problem by taking one course in Machine Learning.
Link to the S&P report which contains their rationale for downgrading, future outlook, etc.
Exactly. It is also one of the reasons why some sites such as reddit start out being genuinely intellectual and as the user base grows, they progressively become just banal - which is what reddit is today.
The study in TFA indicates nothing.
It includes reviews by anyone who has never used the product. Example, this Zenith watch. I won't debate the hilarity of the reviews, but no doubt they are fake.
I think the "undecidability" factor comes into play in any machine learning problem. For example, which email is spam and which isn't? But significant strides have been made to solve that problem and I don't see why online review system is any different.
Are you assuming that all Americans with same qualifications as H1-B visa holders are jobless? I highly doubt that.
The current law stipulates that no more than 65,000 H1-B visas be issued each fiscal year. As of July 18, only 20,500 of these are filled. Of those, around 12,800 hold a Master's degrees. Are you suggesting that stopping the H1-B program is going to improve employment in US? No. What you'll be doing is reducing the quality of workers in American workplaces.
As regards to removing taxation, I am somewhat ambiguous. One could argue that since H1-B holders are not eligible to receive Social Security benefits, so they should not pay taxes However, a significant number of H1-B holders stay on to obtain a green card, making them eligible for the said benefits. Personally, I have no problem paying the Social Security taxes. The job that I am going to start in a month or so is fairly high-tech, the work is very intellectually stimulating and is simply not done in my home country. I am more than happy to help the currently unemployed Americans by paying my share of taxes. I view this as a price I pay for the better living conditions and greater job satisfaction. Mind you, that does not mean that either of them will be able to do my job. If any American is as skilled as I am, he won't have any problem getting a similar job position as mine.
The German movie Das Experiment is based on this experiment. Although they exaggerate quite a lot towards the end, first few days of the movie are real. Overall, an entertaining watch.
I've tried it myself and found it to be horribly buggy/unstable! At least I can work on Superputty without worrying that it'll crash any moment.
My search for a good version of tabbed putty still continues. Currently I'm using Superputty. Unfortunately, it is far from a finished product - the main problems being lack of proper keyboard shortcuts and non-regain of focus on maximize/minimize.
If someone knows a good tabbed putty version, my life would become a lot simpler.
Interestingly, software is affecting law industry big time. Gone are the days when lawyers needed to burn the midnight oil to research relevant cases. With a document managing software with proper tagging features, even you can become a lawyer!.
I think he's being sarcastic.
It was Dana Ballard. IIRC, the work that I referred to was done when he was at Rochester. He seems to have a link to the peanut butter and jelly video up on his page, but I am not able to download it.
I couldn't gather much from TFA apart from that they recognized the activities and the learning model behind it. I don't recall how much training data Dr. Ballard used, but the activity recognition in real time was very accurate.
My Machine Learning professor at University of Texas was hugely into Reinforcement Learning. He and one of his graduate student had developed this kind of system many years ago. Basically, only using a simple video camera, the computer could identify correctly all sub-activities involved in making a peanut butter sandwich; such as 'open jar', 'jelly on bread', 'jar closed', etc. We were shown the video during first class of the course and it was pretty cool.
The MSR guys are doing interesting stuff but this is nothing groundbreaking.