What Windows users?

July 28, 2012 in Internet

TUAW, on Safari 6:

But [Safari] Windows users have noticed something a little strange: There are no download links for a Windows version, and the latest version is still Safari 5.1.7 on Apple's official website.... Whatever the reason, Windows users wanting the new Safari will have a bit longer to wait.

...what Windows users?


Update 8/1: The fix I described below has just been added to the development branch. Mountain Lion users can install the development branch with:

pip install -e git+https://github.com/scipy/scipy#egg=scipy-dev

(note this requires a Fortran compiler; see here for more detail)

I've been updating my post on installing Python/NumPy/SciPy/IPython on Lion to work with Mountain Lion. For the most part, everything works smoothly -- except SciPy.

Trying to compile the stable version of SciPy under Mountain Lion results in errors. It can be fixed by adjusting four files that refer to the deprecated vecLib library and need to be updated withApple's Accelerate framework. A pull request containing the fix has been submitted but not yet accepted into the development branch, and even upon acceptance may take some time to appear in the stable version.

In the meantime, cutting-edge users will need to install SciPy from the modified branch. The following code will do the trick -- at your own risk!!
pip install git+https://github.com/minrk/scipy-1.git@accelerate
Please note that this fix is not guaranteed to be bug-free -- in fact, it will fail some BLAS tests. If you absolutely need a stable version of SciPy, you should consider holding off on upgrading to OSX 10.8 until it is fully supported by the main SciPy branch.


I just finished a clean install of Mountain Lion and have run into what appears to be a major bug: I can't browse my old Time Machine backups!

It seems like I'm not the only one.

This is a major, showstopping bug and it's hard to imagine how it got through Apple's QA process. I hope an update is issued soon.


Caveat emptor

July 7, 2012 in Finance

Another day, another pleasant surprise from the NYT: Tara Siegal Bernard has written a nice article encouraging a degree of educated skepticism when dealing with financial advisors. It's critical to keep in mind that not everyone offering financial advice is bound by any form of fiduciary duty to his or her clients. They aren't even required to disclose their credentials -- or lack thereof. Caveat emptor, indeed!


To mark the iPhone's 5 year anniversary, comScore has released a chart of mobile use by iPhone type:

The most striking thing about the chart, to me, is how steady the width of the bands remains through time. There is a perception that a small but dedicated group of iPhone owners upgrade their hardware with each new release, inflating the number of devices sold (this idea is preposterous in the face of the number of units Apple ships each quarter, but it persists nonetheless). Here, the evidence shows that the release of a new phone draws new buyers rather than cannibalizing old ones. Each strata grows only when its respective device is on sale, which makes sense -- note that both the iPhones 3GS and 4 remained available as low-cost alternatives to the subsequent flagship.

(comScore via Engadget)


One company is doing a fantastic job of rolling out an innovative, attractive, intelligent and bold user interface across its entire product line: Microsoft.

Here's a preview of their Metro-inspired home page redesign. Remember how apple.com looked stuck in the 90's until early last year? It looks positively dated next to this.

For bonus points, try resizing the window...

(Thanks to DF for tipping me off.)


I've just finished Turing's Cathedral, a wonderful new book by George Dyson about John von Neumann's team at Princeton that built one of the first computers. In the title chapter, there are a few excellent quotes:

"I asked [Turing] under what circumstances he would say that a machine is conscious," Jack Good recalled in 1956. "He said that if the machine was liable to punish him for saying otherwise then he would say that it was conscious...."

Jack Good would later explain that "the ultraintelligent machine ... is a machine that believes people cannot think."

One of Turing's key observations is also nicely detailed:

Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child's? Bit by bit one would be able to allow the machine to make more and more 'choices' or 'decisions.' One would eventually find it possible to program it so as to make its behavior the result of a comparatively small number of general principles. When these became sufficiently general, interference would no longer be necessary and the machine would have 'grown up.'

And finally, Dyson sums it up quite well:

The paradox of artificial intelligence is that any system simple enough to be understandable is not complicated enough to behave intelligently, and any system complicated enough to behave intelligently is not simple enough to understand.

(NYT articles notwithstanding, of course.)


My inbox has been buzzing with links to an interesting new research paper from a team at Google led by Andrew Ng (of Stanford AI fame) and Jeff Dean. However, I'm receiving far more links to an NYT piece covering the research. It's great that the work is getting mainstream coverage, but somewhat unfortunate because the NYT has managed to invent spurious details about the research while diminishing its real importance.

So I'm stuck wondering: why has the NYT chosen this moment, this research, this team to finally chime in on my little (but rapidly expanding) research area?

This is hardly the first unsupervised image recognition model, nor the first such model built by Google, and certainly not the first one that can recognize cats! (In fact, you can try such a system yourself, right here.)

Is it possible that we have reached a point where anything related to artificial intelligence -- which in this case I define loosely as "computers doing things that the average person didn't know they could do" -- gets prime billing? Is that a good thing? In the case of the ongoing Big Data trend, such spotlighting has led to a ream of "experts" with marginally more knowledge than a high school statistician and a widespread misperception within the tech community about 1) what Big Data is and 2) what can be done with it. I absolutely, under no circumstances, want machine learning to go down the same path.

We are close enough as it is. The response to the NYT article on Twitter -- millions of people spouting nonsense about "brains" and "singularities" -- demonstrates that while most people clearly have no idea what the research represents, they have strong views about it and, more worrisome, think they understand it (or believe that they should appear to do so). The simple proof, to me, is that most people are linking to the NYT article and not the research itself. Normally, I'd have no problem with that, but the article so grossly distorts the research that it is impossible for anyone who is familiar with the paper to recognize the subject of the article at all.

To get a sense of what I'm talking about, simply turn to page 2 (a very suspiciously-broken page 2, at that) of the NYT article, where Dr. Ng states, "A loose and frankly awful analogy is that our numerical parameters correspond to synapses [in the human brain]." The article's author apparently chose to disregard this wisdom, as the first paragraph of the article describes the system as nothing less than "a model of the human brain." Indeed, the most accurate description is probably that the model represents an abstraction of a certain type of process which likely exists in certain parts of the neocortex.

Moreover, the NYT describes the model as being "turned loose on the internet" where it "looked for cats." I never thought I'd accuse any paper of sensationalism with regard to math, of all things, but here it is! The truth is that the model was fed a steady diet of curated YouTube stills, and brought immense processing power to bear in order to recognize common features in those images -- cats being one of many thousands of categories it recognized. And to be clear, this isn't a classification network per se. Each "neuron" becomes sensitive to a certain combination of patterns and shapes. The researchers went looking for a neuron that was particularly sensitive to humans, and another partial to cats -- the firing of that neuron corresponds to classification of an image. A classifier trained on top of this network would likely outperform even these results.

I'd be as guilty as anyone if I didn't address the most exciting aspects of this new research. First let's separate what's new and what isn't, though. The model itself isn't entirely new. It's an autoencoder, an unsupervised model that builds a representation of its inputs. It has been the subject of much research in the last half-decade, particularly by Marc'Aurelio Ranzato -- first at NYU with LeCun, then at Toronto under Hinton-- whose name I was very happy to see attached to this research. I was worried that after his hiring by Google, we'd never hear from him again, and his research and writing is consistently clear and impressive. More than that, it's a deep autoencoder, meaning an autoencoder builds a representation of the data, which is then fed into a second autoencoder, whose output is passed to yet a third autoencoder... all the way through eight layers in total. This provides a mechanism to aggregate detail from the hyperlocal to global scale, in practice passing on only the most salient features to the next level. This is also not new; deep networks have been closely watched since Hinton's critical work in 2006, and layered networks were introduced by LeCun back in 1989 (with restrictions appropriate for technology of the time).

What is new -- and exciting -- is that unlike LeCun's convolutional networks, the weights of this autoencoder did not have to be tied. This allowed spatial invariance to develop to a greater degree than previously possible: the model was able to recognize pictures which were rotated, skewed or inverted from other examples it had seen. This was only possible because of the incredible amount of resources spent on the project -- untying the weights increases the numbers of parameters by a few orders of magnitude, resulting in a billion free parameters in this case. That result -- and the method by which it was implemented -- is the single greatest advance the research represents.

Students of this field will recognize that there is nothing revolutionary about a network that trained itself, nor one that learned to recognize cats, nor one trained as an autoencoder with SGD. And that brings me back to my original question: why has the NYT decided to write about this model at this time? Maybe it's the Google factor -- decades of research from the finest academic institutions is all very well, but it doesn't have the allure of this company's efforts. If so, I'm sure that would come as a sad realization to many of the Grey Lady's target audience. Maybe we've already crossed the "Big Data" event horizon, where writing about trendy nonsense attracts trendy readers, who in turn enforce the network effects the news is so dependent on.

Or perhaps I'm being too harsh. Shouldn't I be ecstatic to see mainstream coverage of something I spend so much time working on? I don't know. I really feel that misinformation is one of the most dangerous things in the world, and a misinformed public will manage to make worse decisions than an uninformed one. My excitement about seeing this article and its Twitter response rapidly faded when I saw that the people tweeting it either didn't read it, didn't understand it, or didn't bother to try at either one. I don't think this is a case where someone has to go through years of literature to get comfortable with the results, either (though I certainly recommend it!). As I mentioned, the page 1 characterizations of the model were at odds with the researchers' own page 2 descriptions -- that should be enough to make anyone suspicious.

There's a fantastic SNL skit in which Steve Martin is told, repeatedly, "Don't buy stuff you can't afford." I'd like to instate a similar rule for Twitter: Don't tweet things you don't understand. Sadly, I'm informed that would eliminate most of their traffic.

I was shocked to see the following bold headline this morning, as it describes something which did not actually happen:

CBOE 'fear guage' drops 6% in Monday trading

Let's be clear: the VIX is measured in percentage points. A 6% drop in the VIX is very significant, as it represents a 6 point drop in the index level. The drop referenced here was just 1.2 points from yesterday's 20.79 close -- so if the VIX was a stock (or something not measured in percents), it would be accurate to say there was a 6%. But the VIX is measured and expressed in percent, so the change is simply 1.2%. The headline was written either to mislead investors or by someone who doesn't understand the first thing about the numbers they've been tasked with writing about.

How ironic that a headline about the "fear index" is itself sensational! As an investor who looks to sites like MarketWatch for news, I find this absolutely unacceptable -- and it reinforces my (sad) belief that you must always do your own research, and you should never trust anything that anyone tells you.

Even worse, this isn't the first time I've used a "Misreading misleading...: VIX edition" headline. This index is ripe for information manipulation.

(Also, what's with the Facebook "like" button for tickers? Seriously? )


An NYT article about a text-message-based ad that aired during the Super Bowl talks about the high follow-through rate that the ad earned for its creator, the NFL. In fact, the ad did so well that one executive described it like this:

While Mr. Berman [general manager of NFL Digital Media] declined to say exactly how many people went ahead and signed up, he said the number was “exponentially higher” than the 2 percent conversion rate for most Web sites.

What does "exponentially higher" really mean in the context of a single number? Is 4 "exponentially higher" than 2? Is 6? Is 10? Maybe. The trouble is that they are all "linearly higher" as well (yes, I'm making terms up). The problem here is that to describe a pattern or growth as exponential requires a few datapoints. In fact, it even takes multiple observations to characterize something as linear! The most we can say about a single observation is that it is some multiple of another, baseline measure. To say a single number is "exponentially higher" than some other single number doesn't actually have any meaning at all!


{ 1 comment }