prat

If I were you, I would have commented after finishing up with the phone.

One of the issues with NN is that there is no way to find out the relative significance of the input variables and how they affect the output response. Atleast I couldn't find a way when I tried using them a couple of years back. Secondly, I guess there isn't a way to cluster similar cases together to analyze what rules bind different decisions together. Also, one cannot find the partial dependence of the output on a given input variable. All these problems are solved by the random forest classifier which I have been working with and really like.

Thanks for posting this. I wonder if they ever come up with a sequencing platform that relies on millions of nano machines crawling along the dna, sending signals of which nucleotide they just traversed.

"stories" are the tragedy with science. I don't care who was called what and who changed stance. All I need to know is both sides of the argument and the strengths of their respective evidences.

Solar variation and ocean effects are known cyclicals while man made CO2 is monotonically increasing. So the worst case scenario will arise when cyclicals get in phase with the CO2 effect, i.e. all cyclicals get in heat phase. It is premature to discard the constantly increasing CO2 levels just because the ocean cycles are scheduled to cause cooling in the next decade or two.

Maybe Ajay didn't want the spotlight - so what they did was the next best honor by atleast using his real name.

here is the non-paginated link http://www.wired.com/gadgetlab/2009/10/concept-phones-ceatec...

If TV industry ever wants to make that vision true, they would not only have to make TV viewing interactive but also have to provide a button that gives viewers a choice to skip ads.

"If the bank returns your check marked “Insufficient Funds,” you call them and ask if they meant you or them."

Ha..

"iterative and incremental projects" that's it.

If this the only one you have read, then try slaughterhouse 5 too.

"there was no point in the book where I actively felt a desire to know what would happen next"

I agree - at the same time, after reading a couple of his other books, I feel this is the only kind of fiction that I can simply go to a book store and read for 15 min without feeling the need to finish the book (although I do finish it eventually). His style is quite haiku like in the sense that the beginning or end take a back seat. I should mention here that I am no haiku aficionado - so the reference might sound amateurish to someone who knows.

"Give your readers as much information as possible as soon as possible. To hell with suspense. Readers should have such complete understanding of what is going on, where and why, that they could finish the story themselves, should cockroaches eat the last few pages."

After having read most of Vonnegut's books, I can say that this is a big hallmark of his work. Vonnegut has such a microscopic view of events and characters that doesn't exclude even the slightest and faintest of the details or idiosyncrasies. Its refreshing to have that view but sometimes too much - but more than that, it leads to a very special relationship between the author and reader - Its probably just me, but I think of Kurt as a member of my family or karass because of this.

There have been a number of targeted sequencing experiments of breast cancer cells before: using next gen sequencing platforms and microarrays. Comlpete sequencing is by no means a breakthrough. A complete sequencing (3bn letters) has not been tried for a reason - and that reason is that the increase in information content from complete sequencing is minimal - almost zero. infact it will be wasteful exercise to look at the whole genome when you expect to find problem in a handful of genes. Moreover the full sequence of a tumor cell varies from patient to patient and also between cell to cell in the same patient (unlike in healthy cells/patients). that's why no one does full genome sequencing of tumor cells - just targeted sequencing of selected genes.

There is another link that provides slightly more info than just the rank http://www.timeshighereducation.co.uk/Rankings2009-Top200.ht...

Like peer review score, employer score etc. But even then, I think this is more like a voting scheme to obtain scores rather than detailed assessment of each department/ research / paper published / nobel laureates produced/ entrepreneurs produced / innovation quotient etc.

Indian Institute of Technology is not listed? That's a shame. There are 7 institutes and atleast one of them should easily be in top 10. But again that's just my opinion.

This video is old - but I though it was worth watching again after reading the paper. http://www.youtube.com/watch?v=W1czBcnX1Ww

I agree - the question always IS when and not whether:) slowdown has a non-zero probability and will happen - the nontrivial part is predicting whether in a decade or five?

All I could gather from the post is - "great incentives attract great talent produces great software".

"China's inevitable slowdown" - what is the basis of that prediction?

Interesting. But shouldn't the title be something like "Those better at counting pulse are better at decision making". Right now it suggests that if you count before making a decision, it will be a good decision.

"[A]s Elizabeth Warren has argued, the idea that most Americans have been spending frivolously on consumer goods actually isn’t true. Instead, a hefty chunk of the increase in consumption in recent decades has been the result of higher housing prices"

I don't disagree, but past 1-2 decades have seen commoditization of housing, housing and related products started to resemble more and more like consumer goods in the U.S. than in any other country.

some maps in pictures. http://news.bbc.co.uk/2/hi/in_pictures/8284655.stm

Unless there is a contractual clause saying that google employees shall have no services provided to them from another company (and from the word "forces" I assume there is not), google should not have forced this employee. Insecurity is not a hallmark of a genuinely better company.

Here is the publication

http://www.nature.com/nphoton/journal/v3/n10/abs/nphoton.200...

There is a huge and growing R community. I am using randomForest (originally in Fortran) package as a classifier for my dataset.

Here is the detailed article http://www.technologyreview.com/biomedicine/17523/

This is completely new!

That's true. I also think that differences are natural and so is the propensity for like-featured (race/religion etc) people to stick together. But again natural is not necessarily good. But as I write that, I wonder how the global culture (universities, workplace etc.) has helped people to intermingle. Now who is to say that creation of this global culture is not natural. I think it is. What's man made is natural too. So I take back what I said about natural not being good - natural today is different and better than natural yesterday.

Original Nature article http://www.nature.com/nature/journal/v461/n7263/full/nature0...