[ Professional | Publications | Personal | Lab | Notes]
School of Electrical Engineering and Computer Science
Oregon State University
Note: The best way to reach me is by email, not by phone.
- Office: Kelley 2075
- Email: my last name at eecs dot oregonstate dot edu
- Mailing Address:
- School of Electrical Engineering and Computer Science
Oregon State University
1148 Kelley Engineering Center
Corvallis, OR 97331-5501
At the Forth Rail Bridge (Edinburgh, Scotland)
(Sept 23, 2020) I'm the OSU Site Director for the Pervasive Personalized Intelligence Center, which is a National Science Foundation funded Industry-University Collaborative Research Center. If you are from industry and interested in collaborating with us, please send me an email!
(Sept 23, 2020) I'm now a Full Professor (quite a contrast to the days of being a hungry grad student)
(Oct 1, 2018) I am back at Oregon State University after serving as a Program Director in CISE/IIS/RI at the National Science Foundation.
- (General) Anomaly Detection
Anomaly detection is a very vaguely defined area of machine learning / data mining but at a high-level, it involves discovering unusual but meaningful data instances. I am interested in anomaly detection for structured data, explaining anomalies and harnessing user feedback to improve anomaly detection.
- Open set recognition
Open set recognition requires a classifier to accurately recognize classes seen during training and also detect when new classes that were not seen during training. This problem is closely related to out-of-distribution detection and plays an important role in making machine learning robust to real world problems.
- Rare category detection
A challenging problem in anomaly detection is to discover interesting anomalies and not just statistically significant ones. Rare category detection is a human-in-the-loop process that bears a resemblance to active learning. In rare category detection, the machine learning algorithm presents representative instances of each class to a user for labeling so that the user can discover as many classes as possible given a budget of queries.
Explainable AI (XAI)
- Explaining supervised learning
I am interested in explaining deep learning classifiers, with a focus on understanding what is going in latent representation layers. In addition, I am interested in developing techniques to produce counterfactual explanations to help users of machine learning algorithms understand thebehavior of their algorithm.
- Species Distribution Modeling
Species distribution mapping involves predicting the occurrence of a species from a set of environmental features (such as temperature, precipitation, vegetation, land use, etc.) Species distribution modeling can be viewed as a supervised learning problem, but a large number of real-world problems make it much more challenging. One major issue I am working on involves dealing with imperfect detection of species, which is a problem encountered in large biodiversity citizen science datasets like eBird .
Temporal, Spatial and Spatio-temporal Machine Learning
- My research in this area includes weakly-supervised activity recognition from time series data, spatial anomaly detection and discovering latent spatio-temporal patterns.
Deep Explainable Singularity Detection via Inverse Mandalorian Monte Carlo
- I don't really do research in this non-existent area but this paragraph is here to weed out graduate applicants that blindly cut and paste from my research interests.
- IJCAI 2021(Aug 21-26, 2021), Montreal, Canada: Abstract submission Jan 13, 2021, Full paper Jan 20, 2021.
- ICML 2021(July 18 - 24, 2021), Virtual: Abstract submission Jan 28, 2021, Full paper Feb 4, 2021.
- UAI 2021(July 27 - 29, 2021), Virtual: Full paper Feb 19, 2021.
- KDD 2021(August 14 - 18, 2021), Singapore: Full paper Feb 8, 2021.