Research
I am interested in learning and inference with probabilistic graphical
models, especially combining learning and inference with arithmetic
circuits and combining first-order logic with uncertainty in
statistical relational AI. I have also worked on desktop activity
recognition, spam filtering, and recommender systems.
Learning for Efficient Inference
Inference in Bayesian networks and Markov networks is intractable in
general, but many special cases are tractable. Often, a tractable
subclass such as naive Bayes mixture models yields comparable accuracy
while offering exponentially faster inference (
Lowd and Domingos,
2005
[
pdf]
[
ppt]
[
appendix]).
Furthermore, by incorporating a preference for tractable
models into the learning algorithm, we can guarantee efficient
inference without restricting ourselves to any particular class (
Lowd
and Domingos, 2008
[
pdf]
[
pdf+proofs]
[
ppt]).
Given an intractable model, we can use learning methods to find an
accurate but tractable approximation to the original (
Lowd and
Domingos, 2010 [
pdf]
[
proofs]
Software:
- Libra toolkit --
Exact and approximate inference for BNs and MNs,
BN structure learning, and more.
- NBE --
Efficient probability estimation using mixture models.
Statistical Relational Learning
Statistical relational learning seeks to represent the complexity and
uncertainty present in most real-world problems by combining
first-order logic with probability. The main challenges are in
developing effective representations and effective algorithms. One of
my projects has been Recursive Random Fields (RRFs), a multi-layer
generalization of Markov logic networks that resolves a number of
inconsistencies in the Markov logic representation (
Lowd and Domingos,
2007a
[
pdf]
[
ppt]
[
ppt+audio]).
I have also worked on applying quadratic optimization
algorithms to Markov logic weight learning, resulting in more accurate
models in much less time than before (
Lowd and Domingos, 2007b
[
pdf]
[
ppt]
[
video]).
Adversarial Machine Learning
I spent the summer of 2004 at Microsoft Research working with
Chris Meek on
the problem of spam. We looked at a common technique spammers use to defeat
filters: adding "good words" to their emails. We developed techniques
for evaluating the robustness of spam filters, as well as a theoretical
framework for the general problem of learning to defeat a classifier
(Lowd and Meek, 2005ab
[
pdf]
[
pdf]).
Slides from a talk at Oregon State University
(7/14/2006).
Slides from a talk at the 2007 NIPS
Workshop on Machine Learning in Adversarial Environments for Computer
Security (12/8/2007).
Slides from a talk at the
University of Cagliari, Italy (7/3/2008).
Slides from a talk at Portland State
University (11/30/2009).