I am a postdoctoral scholar at the University of California, San Diego
under the Data Science Postdoctoral Fellowship
program, which is co-sponsored by ITA
, the Qualcomm Institute
. My work is in the area of statistical machine learning. Previously, I was a postdoctoral scholar in Lise Getoor
research group at UCSC
, and I graduated from Padhraic Smyth
group at UCI
. For more information, see:
- My publications, including links to code and slides.
- My Courses.
- My CV (or email for a more up to date version).
- My PhD thesis has a lot of tutorial-style introductory material on probabilistic latent variable modeling, which you may find useful.
My research interests are in both applied and foundational machine learning
, focusing on probabilistic models
and Bayesian inference
algorithms to learn them from data. My work aims to promote the practice of latent variable modeling
for applied research in disciplines such as computational social science
and the digital humanities
. I develop models and algorithms for finding interpretable latent structure in data such as social networks and text, which may provide a window into human nature.
- I have accepted a tenure-track position at the University of Maryland, Baltimore County (UMBC), Information Systems Department, beginning Fall 2017!
- Our paper, Dense Distributions from Sparse Samples: Improved Gibbs Sampling Parameter Estimators for LDA, was accepted at JMLR! See the ArXiv preprint.
- New preprint uploaded to the arXiv: Mixed Membership Word Embeddings for Computational Social Science.
- I am teaching CSE291D Latent Variable Models again in Spring 2017. See last year's course webpage.
- Our paper on differentially private EM was accepted at AISTATS!
- I gave a talk at the UCSD AI seminar (Winter 2017).
- I am a co-chair for the 2017 Information Theory and Applications Workshop.
- Winner of the SoCal Machine Learning Symposium 2016 runner-up prize for best presentation!
- SoCal Machine Learning Symposium extended abstract on mixed membership word embeddings accepted for oral presentation!
- New NIPS workshop paper on privacy-preserving topic modeling.
- I co-presented a tutorial on generative models for social network data at SBP-BRiMS 2016 with Kevin Xu. [Slides]
- New preprint on practical privacy for EM on arXiv.
- Our paper on privacy-preserving Bayesian inference was accepted to UAI.
- In Spring 2016, I taught an advanced graduate-level course in probabilistic machine learning, CSE291D Latent Variable Models.
- I was a workshop co-chair for the 2016 Information Theory and Applications Workshop.
- I accepted a postdoctoral scholar position at UCSD, affiliated with Calit2, ITA, and CSE.
- I gave 5 oral presentations at 2015 summer conferences: ICML, KDD, RecSys and ACL (x2).
- See publications for links to software associated with my papers.
- At the behest of the University of Waikato Department of Mathematics, I created Tuatara Turing Machine Simulator , a graphical Turing machine simulator and construction tool for teaching purposes.
- I also worked on the GUI for the Boundary Visualizer, a classification visualization tool in WEKA , a popular java open source data mining toolkit developed at the University of Waikato.
- I was a member of the Japanese taiko drumming groups Watsonville Taiko and Waitaiko , and the Korean drumming group Hansori at UC Irvine.
- I played guitar in the rock group 4 Second Fuse.
- In 2007-2008, I was part of the problem reviewing/writing team for the ACM South Pacific Programming Contest.
- For a number of years, I was involved with the executive committee of the Waikato ACM Student Chapter .
- My Erdös number is two (James R. Foulds - Leslie R. Foulds - Paul Erdös).