Rob Hall
I am now a programmer at Etsy

My thesis is here.

I was a grad student in the Machine Learning department at CMU. I was working with Professor Steven Fienberg of the statistics department, looking at ways to do some traditional data mining tasks (density estimation, classification etc.) in a way which respects ``privacy'' of the underlying data.

Previously I worked with Andrew McCallum at UMass, looking at machine learning for record linkage in a variety of different settings.

Before that, when I was an undergrad I did a project with Arnold Rosenberg regarding grid computing whereupon I lowered my Erdos number to 3.




Differential Privacy for Functions and Functional Data
Rob Hall, Alessandro Rinaldo, Larry Wasserman


Valid Statistical Inference on Automatically Matched Files
Rob Hall and Stephen Fienberg
PSD 2012

Achieving Both Valid and Secure Logistic Regression Analysis on Aggregated Data from Different Private Sources
Stephen Fienberg, Robert Hall, Yuval Nardi
Journal of Privacy and Confidentiality, 4(1), pp 189--691.

Differential Privacy in Reproducing Kernel Hilbert Spaces
Rob Hall, Alessandro Rinaldo, Larry Wasserman
ICML 2012 Workshop on Reproducing Kernel Hilbert Spaces.


Random Differential Privacy
Rob Hall, Alessandro Rinaldo, Larry Wasserman
Under review.

Secure Multiparty Linear Regression Based on Homomorphic Encryption
Robert Hall, Stephen Fienberg, Yuval Nardi
Journal of Official Statistics, 27(4), pp 669 -- 691.

Approaches to Multiple Record Linkage
Mauricio Sadinle, Robert Hall, Stephen Fienberg
ISI 2011, invited paper


Privacy Preserving Record Linkage
Robert Hall, Stephen Fienberg
Privacy in Statistical databases (PSD) 2010


Translating Relevance Scores to Probabilities for Contextual Advertising
Deepak Agarwal, Evgeniy Gabrilovich, Robert Hall, Vanja Josifovski
CIKM 2009
slides (rough).


Unsupervised Deduplication using Cross-Field Dependencies
Robert Hall, Charles Sutton, Andrew McCallum
KDD 2008


Canonicalization of Database Records using Adaptive Similarity Measures
Aron Culotta, Michael Wick, Robert Hall, Matthew Marzilli, Andrew McCallum
KDD, 2007

Author Disambiguation using Error-driven Machine Learning with a Ranking Loss Function
Aron Culotta, Pallika Kanani, Robert Hall, Michael Wick, Andrew McCallum

First-Order Probabilistic Models for Coreference Resolution
Aron Culotta, Michael Wick, Robert Hall, Andrew McCallum

A Comparison of Dag-Scheduling Strategies for Internet-Based Computing
R. Hall, A. Rosenberg, A. Venkataramani.
IPDPS 2007


Thesis defense
ppt pdf
Nov 5th, 2012.
Secure Multiparty Regression Based on Homomorphic Encryption
Penn State University, Theory Seminar March 2011

Arbitrary Other Stuff

My Erdos Number is 3 due to collaboration with Arny Rosenberg (2) during my final year as an undergrad at UMass.

In addition to machine learning I also enjoy running and have begun tracking my progress here

During the summer of 2012 I have rediscovered my interest in writing little openGL demos, which is now made even simpler due to the advent of webGL you can see some little demos I've made here.