Kenny Shirley
Principal Member, Technical Staff
Statistics Research Department
AT&T Labs Research

I am a member of the Statistics Research Department at AT&T Labs in New York City, where I work on hierarchical Bayesian modeling, MCMC methods, visualization of hierarchies, text mining, and other topics related to applied statistics.



Last night I gave a talk for the NYC Sports Analytics Meetup Group, and it was a blast! There were lots of great sports researchers and enthusiasts in the crowd. My talk was about Baseball Hall of Fame voting, of course. Here is a link to the slides from my talk.

1/13/2014: (2-part news!)

Part 1: Last month I moved to our new NYC office with about 20 colleagues. (The rest of our lab moved from Florham Park, NJ to new office space in Bedminster, NJ). Our NYC office is a newly renovated space at 33 Thomas Street in Tribeca. The building is pictured below -- let's just say it's very secure, and hard to miss if you're walking around the neighborhood. I'm hoping we can start hosting some talks and workshops to get involved in the NYC tech scene.

Part 2: I've recently become slightly obsessed with Baseball Hall of Fame voting. After our interactive visualization of historical voting was featured on Deadspin as one of the 12 best sports infographics of 2013, I figured the next step would be to fit a model to historical data to predict Hall of Fame voting. Here's a link to my analysis and results. The 2014 predictions, pictured below, weren't great; we did OK with Maddux and Biggio, and pretty poorly with all the rest of the candidates! But for 2015 I like the initial predictions: Randy Johnson and Pedro Martinez are locks to get in, and John Smoltz is borderline. I'm planning to re-visit this throughout the year to improve the model.


My paper with Amy Reibman and Chao Tian was recently accepted to the IEEE International Conference on Image Processing (aka ICIP). The paper is called "A Probabilistic Pairwise-Preference Predictor For Image Quality", and in it we describe a multilevel Bayesian model that can be used as an objective image quality estimator. To gather data, we used Mechanical Turk to run an experiment in which subjects viewed pairs of images online and indicated which image from each pair they felt was of higher quality. The images were systematically degraded with various types of distortions at various levels of severity. From our model we inferred the effects of the distortion types, their severities, and existing objective quality estimators, while controlling for the effects of subject-specific bias (where subjects systematically tend to prefer either the left or right image for some reason, all else held equal) and reference image bias (where subjects tended to prefer the image of the elephant compared to the image of the barn, for example). For all the details, here's the pdf; Amy will be presenting it at the conference in Sydney next month.

For older news, click here.