18 Nov 2017

Well, here it is, my new website. It only seemed fitting, next to my new job in a new city, my new life after graduate school. I will be updating this space regularly now that I do not have the specter of a dissertation looming over me. Hello again, World.

03 Oct 2015

The first-ever spacewalk was performed by cosmonaut Alexei Leonov on March 18, 1965. Since then, humankind has logged over 2,000 hours of extravehicular activity (EVA) beyond Earth’s appreciable atmosphere. Three nations have led spacewalks: Russia, USA, and, most recently, China, following the success of Shenzhou 7 on September 27, 2008.

I developed the following graphic after stumbling upon NASA’s data portal and their dataset on US and Russia’s Extra-vehicular Activity (EVA). The processing and visualization of these data are fully reproducible at github.com/nsgrantham/spacewalks.

More …
18 Sep 2015

I review the history of deep learning from perceptrons to convolutional neural networks and briefly discuss its recent successes in machine learning. Watch the videos below and follow along with my slides.

More …
04 Feb 2015

Introductory calculus courses often give students the following rules-of-thumb when finding limits of ratios of functions:

- Exponential \(\exp(x)\) beats a power \(x^{k}\)
- Power beats a logarithm \(\log(x)\)

For example, this implies \(\lim\limits_{x\to\infty} \frac{\exp(x)}{x^{2} + 7x + 42} = \infty\) or \(\lim\limits_{x\to\infty} \frac{\log(x)}{x^{4}+x-12} = 0\).

However, courses tend not to emphasize just how ridiculous the factorial function is in comparison. Recall that the factorial of a non-negative integer \(n\) is defined such that
\[
n! = n\cdot(n-1)\cdots 2\cdot 1,
\]
where \(0! = 1\) by convention. So, say \(n = 4\). Then \(n! = 4\cdot 3\cdot 2\cdot 1 = 24\).

Factorials are absolute monstrosities for even moderately-sized \(n\). To appreciate just how quickly factorials grow, let’s consider looking back \(n!\) seconds in the past.

More …
14 Jan 2015

Pitchfork is the largest indie music site on the Internet (in the English-speaking world, at least), updating its pages daily with the latest indie music rumblings, interviews with budding artists, sneak previews of new albums and artist collaborations, and, most notably, a suite of music reviews by dedicated music critics forming Pitchfork’s staff. I follow Pitchfork’s album reviews religiously and I am not alone in feeling that their ‘Best New Music’ category routinely captures the best that modern music has to offer.

Since its creation in 1999, Pitchfork has gained quite the following among music fanatics and its widely-read album reviews can have a demonstrable impact on an album’s success. Indeed, Pitchfork is credited with propelling Arcade Fire into the limelight after bestowing a glowing 9.7 ‘Best New Music’ review on their 2004 release *Funeral*. Conversely, they are responsible for striking a devastating blow to Travis Morrison’s solo career after his 2004 release *Travistan* received a controversial 0.0. This make-or-break phenomenon has appropriately been dubbed The Pitchfork Effect.

Simply put, an album’s commercial fate can rest squarely in the hands of the Pitchfork staff member responsible for its review. But what goes into a review? Are staff members consistent in their reviewing behavior? Do biases exist? If so, what kinds? In the following post, I analyze Pitchfork’s data in an attempt to answer these questions. After all, who reviews the reviewers?

More …