Has anyone else noticed that in all aspects of every day life that involves speaking, that “theres” is a seemingly perfect substitute for “there are?”

thatisall,

-Derek

Yes, the postings are infrequent. We’re all a little too awesome when we’re not on the internet, so forgive us. Moving on!

This might become a new series here. I encourage all other posters to jump on board with this one. It’s called “If I Ever:…”, where you explain what would happen if you ever have to encounter something again. Let’s see an example:

If I Ever:

Experience: Have to explain how to attach a text file in an EMail to a college senior between the ages of 19 and 21…

Solution: I will poke them in the eyes.

Seriously. College seniors. Digital natives. Can’t attach files.

someweren’tsurehowtogettotheiremailinthefirstplace,
-Derek

As a grad student – food needs to be fast, and delicious. The whole nutritious part is questionable.

I’d like to share my newly perfected grilled cheese recipe. It knocks the spots right off the cows that produce the cheese.

Required Cheeses:

• Sharp Cheddar
• Blue cheese (real, and real good)
• Fontina (the dry type, almost like an Asiago)

Required Spices:

• Garlic Powder
• Dill Weed
• Black Pepper

And the secret to making this the greatest sandwich in history? Use peanut oil instead of olive or vegetable oil, or butter. I’m not a very big fan of peanut oil – but it makes grilled cheese sandwiches magical. I’m considering just putting peanut butter directly on the sandwich.

ohmanthatwastasty,
-Derek

Stop being non-sexy. Give up on Word, OpenOffice.org, Pages, LaTeX and any other word processing environment you can think of, except one: Google Docs.

Before I begin: I hope James chimes in on this with some of his material regarding Zotero and Google Scholar!

So we’ve had some posts flouting the greatness of Zotero and Google Scholar; but apparently we’re not even close to scratching the surface of the absolutely mind-blowing greatness of Zotero, or the fantastic awesomeness of Google Docs.

Through some of my most recent digging, I’ve found out that Google Docs supports LaTeX for equation editing. That’s awesome, because LaTeX is the cleaniest, most ubiquitous way of doing equation type setting.

But how about we try something else on, while we’re in the lingerie shop of academic writing: You can drag and drop Zotero references into Google Docs and it just works. AWESOME!

And let’s not forget  that Google Docs allows us to roll back our versions, whilst seeing who edited what and how, just like a revision system for writing code.

i’mstillabitscaredofthecloudthough,
-Derek

Let’s have some fun! We’re going to play a little game. You have to make a guess where and when these phrases came from:

THIS MAN is wanted for treasonous activities against the United States:

1. Betraying the Constitution (which he swore to uphold)
2. He has been WRONG on innumerable issues affecting the security of the US
3. He has appointed Anti-Christians to Federal Office:
4. He has been caught in fantastic LIES to the American people

You’ve probably got some ideas, don’t you? Let’s keep playing! This time, try to fill in the [...BLANK...]

MR. [...BLANK...], despite contentions on the part of your administration, the State Department, [...], and members of your party, we free-thinking and America-thinking citizens of [...BLANK...] still have, through a Constitution largely ignored by you, the right to address our grievances, to question you, to disagree with you, and to criticize you.

In asserting this constitutional right, we wish to ask you publicly the following questions — indeed, questions of paramount importance and interest to all free peoples everywhere — which we trust you will answer… in public without sophistry.

In case you haven’t noticed, these games are related! Now we’re going to play switcho-changeo by replacing some words in the original text with something else! This round of switcho-changeo is directly related to fill in the [...BLANK...]:

WHY have you ordered or permitted your  friend Janet Napolitiano, the Secretary of the Department of Homeland Security, to go soft on Socialists, fellow-travelers, and ultra-leftists in America, while permitting her to persecute loyal Americans who criticize you, your administration and your leadership?

HAVE YOU GUESSED IT YET?!

I bet you have. But I’ll just cram the answers in your face right now:

Game 1:

The handbill distributed the day before JFKs Dallas visit, via Wikipedia

Game 2 & 3:

A full page advertisement the day of the assassination, via UGA Law School

Huh. I think this sounds familiar? I suppose that the Kennedy flyers here, aren’t in any way completely fucking similar to anything going on now. Maybe I think I’m just seeing things?

Crazy loves to repeat itself. I’m assuming, mostly out of lack of originality.

andthatfolksisademonstrationincrazy’slackofcreativity,
-Derek

PS: But seriously – this demonstration of similarity between the levels of crazy is strikingly disturbing.

Edit: So what’s the deal with this post? Over the weekend, I had visited the JFK Sixth Floor Museum, which I find to be one of the most interesting historical exhibits I’ve ever been to, and a very emotional one. I find it disturbing and sad that in 46 years crazy is still completely and utterly crazy. The same vile tactics are used.

• Take whatever isn’t popular and claim that’s how the President rolls. (i.e. exchange Catholic for Muslim, and Communist for Socialist)
• Claim you just want “the truth”
• Claim no party or political affiliation
• Make ridiculous signs that spout crazy like a waterfall

It’s the whole idea that America’s principles are all about the individual. Somehow it’s patriotic to not want social programs for your fellow citizens ?

I realized after the post that it seemed a bit weird and cryptic. BUT FUN! DON’T FORGET WE PLAYED CHALLENGING WORD GAMES LIKE IN THE HIT BOARD GAME CRANIUM!

It’s not a breakdown in the sense that I’m punching random undergrads, nor is it the part of the midterm where we all sing the chorus to Twisted Sister’s “We’re Not Gonna Take It” – you know, that part without music, just vocals?

This is more of a debrief. For myself, and hopefully for others.

My first half-semester as a PhD student has been exciting, interesting, and tiring. Nearly habitually now, my Sundays are the closest nights of sleep, I believe, that I’ve come to insomnia. Is it because of nerves? Stress? Excitement? Possibly all, but probably none. I don’t know, I don’t fall asleep quickly, anyways.

For example, last night I slept the combined total of 1 hour. I woke up at one point (after about 10 minutes) convinced a zombie was gently biting my hand – the key here is gently. It turns out my hand had gone all pins-and-needles before I could go into a deep sleep. I suppose that’s kind of badass. I woke up confused about not seeing a zombie. Maybe next Sunday?

At this midpoint, though, I feel as though I’ve hit a (an intellectual or exhaustion) wall. Though, in part, I’m guessing that (blaming it on) my cognitive psychology midterm. It was a soul-shattering exam. So far, we’ve covered 4 topics:

1. Perception
2. Attention
3. Memory (working and long)
4. Executive function

From 4 perspectives:

1. fMRI/PET
2. EEG (ERP)
3. Normal behavioral – humans and animals (which is a good majority…)
4. (Brain) Lesioned behavioral – humans and animals

Before I  recreate the first question on the test, let me be clear, that I try to see “the bigger picture”, and I think I’m relatively good at it. There is one major theme in our book that seems (to me) to be explicit (but is probably implicit) is that everything up until this point is all about competitive learning. A bigger discussion for a different day.

First question:

Discuss the research and evidence from (a) lesioned patients, (b) fMRI/PET and (c) ERP that supports what we know about perception, attention, memory, and executive function.

Oh you mean everything. In a 3 hour test, it took me 2 hours to answer that question – and I was rushing. I got 3.5 out of 4 points on that question. I even cited outside resources and put them in APA style format – from memory. FFFFFUUUUU!

Well then. I am making progress, overall, and I think it’s good. My first year project is almost sorted out. And by almost, I mean that all of my proposals have been shot down – but for exceptionally good reasons. Everything I’ve proposed so far would take several years and a sizable team, with lots of money. At least I know exactly what I’m not doing.

Combine those with the fact that I ate what must have been a stick of deep fried butter at the State Fair this weekend, and I’ve hit my limit. I need a refreshing start.

Fortunately, two friends came down this past weekend, and we went to the Kennedy museum (amongst other beer related things), where I ranted on like a lunatic about how improbable I think it was for there to be a single shooter in the 6th floor window.

And the posts should come a bit more frequently, now. I’m going to try to make sure that any down time I experience, or lack of motivation to do real work, gets diverted to teh-inter-blogz.

mysoulisstillshatteredfromthattestthough,
-Derek

Originally, I was working on a longer version of explaining PCA. Basically, how I grew to understand it, and when the revelation hit. But I noticed that PCA can literally be computed in two lines of MatLab or R code (one if you’re savvy!), and I can effectively demonstrate the idea much clearer by being short. I will do the long version, though, I promise.

The whole point of PCA is to find a new set of axes in your data set that are descriptive of some relationship that is not demonstrated by your current set of axes.

Let’s define a few things first:

• Observations (in rows) – Data rows, usually a person or single source of data, for example a person filling out a survey.

• Variables (a.k.a. dimensions; in columns) – All the variables used to define an observation; e.g. each question of the survey is considered a different variable (or dimension). I like using dimension, so we’re going to use that.

• Centering (related terms: Center of Mass, Centroid, Center of Gravity, Barycenter) – Moving the entire data set to the “new” origin, a.k.a. the Center of Mass, Centroid, Center of Gravity, or Barycenter. My favorite term is Centroid.

And, as a preliminary to the explanation, I’m going to tell you why (I believe) it should be called Principal Component Analysis, and not Principal Components Analysis. There is only 1 vector that is in fact, the principal component. All other components are consequent to the first, because of maintaining orthogonality. You’ll see.

Do note, however, there is a huge exception to the idea of 1 component when you start doing weird stuff, like screwing with orthogonality.

Explanation (in words):

Eigen-decomposition == Singular Value Decomposition == Principal Component Analysis.

That’s it. When doing an eigen-decomposition, let’s say calling the eig() function in your favorite environment (MatLab, R) you get back two things: eigenvectors, and eigenvalues.

The eigenvector with the largest eigenvalue tells you where the most variation is going on with the variables. This is your principal component. Your second (principal) component? The eigenvector with the next largest eigenvalue. All components are orthogonal, and that also means that all components are independent.

But what about singular value decomposition? When using your favorite statistical environment you’ll see something like [P,D,Q] = svd(data). P is the same eigenvectors as described above. D are your eigenvalues. We’ll ignore Q for now.

When you report results from PCA, people want 3 things

1. Factor Scores
2. Graphs of components
3. What the components mean

Your factor scores are easy, they are just the eigenvectors of the observations (P) multiplied by the eigenvalues (D). Your graphs are easily attained by multiplying your original data by the eigenvectors of the variables (Q). Number 3 is harder, though. It takes a visual representation.

Major points:

• The order of importance (variance) of your principal components is defined by the eigenvalues in an eigen-(or singular value) decomposition. Sort largest to smallest.
• Components are independent because they are orthogonal.
• You need to visualize to interpret.

Code (for MatLab):

data = rand(100,2);
[P,D,Q] = svd(data);
factorScores = P * D;
rotatedData = data * Q;
figure
plot(rotatedData(:,1),rotatedData(:,2),’b.’)
axis square;

I’ll be following up with a more detailed version of this. Getting to into the nitty-gritty, and the historical context of PCA, as well as the use and misuse of all the terms that are in some way associated to PCA.

thatwaseasy,
-Derek

I was reading an article about a new (possible) vaccination that had (apparent) success for HIV prevention. That sounds really promising right?

Except, this article sort of exploits the ignorance of either the journalists, or whomever is doing the first pass at analysis. I’m no expert — yet, give me until the end of the semester — on analysis like this, but there are some things I do know. Like how to count and what proportions do, and shit.

I’m going to point a few things out, mostly to generate discussion, rather than picking on journalists or whatever.

First, I noticed that the article indicates something about a lower infection rate:

The results: New infections occurred in 51 of the 8,197 given vaccine and in 74 of the 8,198 who received dummy shots. That worked out to a 31 percent lower risk of infection for the vaccine group. Two of the infected participants who received the placebo died.

They work out their 31% lower infection rate by doing this:

1-(51 / 74) = 1-(~0.68) = ~0.31

That’s not very sciency or statisticy. But, let’s go in a different direction for a moment… What is the infection rate of the population, now? According to AVERT, in 1991 there were 143,000 new cases. In 2003, there were 19,000 cases.

OK… Population in 1991 was 56,814,069. Population in ~63,100,000 (Google told me this). So that works out to an infection rate of…

143,000/56,814,059 = ~ 0.002 or 0.2% of the population (infected in 1991)

19,000/63,100,000 = ~0.0003 or 0.03% of the population (infected in 2003)

2003 was quite small. Let’s pretend that the study mentioned above used a correct sampling population, which should reflect actual infection rates.

51/8197 = ~0.006 = ~0.6%

74/8198 = ~0.009 = ~0.9%

The infection rates here are larger than the 1991 infection rate. Could the vaccination population have believed that they are immune, effectively going out and exhibiting riskier sexual behavior?

This analysis, alone, is actually not good. Both populations, as in random, but demonstrative samples, of the population are exhibiting apparently higher infection rates.

Now, there are plenty of methods one can use to see if your random sampled population is in fact reflective of the real population (via techniques like bootstrapping). But, with a cursory look – it doesn’t seem like this was done.

If I find some time tonight, I’ll simulate the results of bootstrapping to show how to (more) genuinely demonstrate the infection rate.

Thoughts?

journalismneedstobemorecareful,
-Derek

Here’s something funny for you: According to Texas, Neil Armstrong is a dickhead. Space is where god lives. NOT SOME DUDE WHO PLAYED THE TRUMPET!

NASA SPENDS ALL OF OUR TEAPARTY TAX DOLLARS ON SENDING MUSICIANS INTO SPACE TO MAKE GOD ANGRY.

I’m sorry. I just needed to vent a bit of stupid.

So. James had a neat post on Zotero tips, which have been hugely beneficial. But, when their powers combine, Zotero and Google Scholar form CAPTAIN PLANET.

No, that other thing,… a robust and friendly research tool.

So this might be simple o trivial, but it has been HUGE for me when I run into one of two situations:

• I found the .pdf on the author’s website, but not exactly the journal submission (no one click citation recording)
• I found the journal article page, and the abstract, but no PDF

Google Scholar provides some neat options that allow you to immediately export the citation from the search page, and also to link the search results directly to your campus library (or any other library of your choice).

In Google Scholar (whilst logged into your Google account), go to Scholar Preferences and:

• Set Bibliography Manager to RefMan ( I find this works best with Zotero )
• Add current campus, or open access library to your Library Links section.

The first one is pretty straight forward, but the second is deceptively more awesome than it seems. Let’s check out a screenshot…

So, you can see that I’m busy researching dinosaurs. But not just any kind, may I add. This is science. It’s rigorous. You’ll note that I can “Import into RefMan” – easy, like I said. But that additional link to the right of the first hit is AWESOME:

That’s right. It immediately searches the UT Dallas library. It can tell me where exactly in the library it is, and if there is no hard copy, exactly where and how to get an electronic copy, or, a copy via inter-library loan.

SCIENCE!
-Derek

I’m not crazy. I told you a while back that religion (as in the crazies, not the regulars) and “intelligent design”/creationism are creeping into computer science, and by what seems to be a clear extension – cognitive science.

Newest example: Uncommon Descent telling us Von Neumann was an ID proponent. I cannot confirm, nor deny this. What I can confirm, however, is that Von Neumann’s work in absolutely no f-ing way supports creation. A pair of quotes from the post I linked above:

In the late 1940’s von Neumann studied from a theoretical point of view the problem of self-reproducing automata (see “Theory of Self-Reproducing Automata”, 1966, University of Illinois Press, Urbana). He was the first to provide an algorithmic model of a self-reproducing automaton. Roughly speaking, he proved that to achieve the goal it is necessary to resolve four problems: (1) to store instructions in a memory; (2) to duplicate these instructions; (3) to implement an automatic factory (sort of “universal constructor”), able to read the memory instructions, and, based on them, to construct the components of the system; (4) to manage all these functions my means of a central control unit. A self-reproducing system must contain the program of its own construction. This program is a sort of consistent and complete abstract image of the system. It is not for chance that the inventor of the theory of cellular automata is also one of the fathers of computer science. In fact, in two words, what von Neumann proved is that self-reproduction needs programming and processors (i.e. software and hardware). The following simplified symbolic picture can give an idea of the von Neumann’s solution:

[...]
Of course the above list of four functions is simplified and reduced at minimum. However without all these functions any self-reproducing automaton cannot work. In the terms of Intelligent Design theory that means that the set {1,2,3,4} is irreducibly complex (IC) at the functional level. Von Neumann’s self-replicating architecture is science at its best and represents an intelligent design. What was his scientific forecast about biological cells but an ID prediction? Given that such ID prediction was proved true in the lab some time after, on that occasion ID was indeed science and the opposite of a science stopper (the usual accusations against ID are: “ID is not science”, “ID is science stopper” …).

And this is really starting to boil my blood. This specific paper is considered the 49th most influential work in the cognitive sciences (prior to 2000). I’m not crazy in thinking that ID and creationist lunatics are intentionally moving their efforts to the cognitive sciences.

Let’s look at the trend I’ve observed since April:

• Case 1, documented at my old-blog-watering-hole, Dear Something.

This was a long, misinformed, and boring article claiming the Turing Test was nonsense. It was an attempt at explaining why minds are not like computers – one of, if not the best, analogy so far to explain how the brain and mind work. The article felt like an attempt at subverting cognitive science and just leaving the mind to philosophical or ideological interpretation.

• Case 2, I rant about it here (in two parts [2]), but provide links to experts who run blogs, too.

This was a strange, and bold move by William Dembski and Robert Marks to literally infiltrate/hijack the IEEE SMC journal and SMC 2009 conference with intelligent design based “research”, effectively claiming that genetic algorithms don’t work as well as we think do; erego – god made people.