Weng-Keen Wong

Assistant Professor
School of Electrical Engineering and Computer Science
Oregon State University

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

Note: The best way to reach me is by email, not by phone.

At the Arc de Triomphe, Paris, France

[ Professional | Research | Personal | Lab | Notes]

News

Daniel Neill and I are giving a tutorial on Event Detection at KDD 2009. Click here for the slides.

Research

My research can be divided into two general areas:

Interactive Machine Learning

I'm interested in machine learning algorithms that require interaction with human beings, such as an interactive process before learning occurs or through a human-in-the-loop process during learning.

Learning User Preferences / Learning from demonstration with Prasad Tadepalli, Tom Dietterich, Ron Metoyer
As part of DARPA's Integrated Learning project, we are interested in learning to perform a complex task from demonstration. We are looking at ways of leveraging different types of knowledge sources to improve performance on a task, especially if certain knowledge sources are limited, expensive to obtain or simply not available. For example, if a large amount of training data is not available, can a machine learn from other forms of inexpensively obtained knowledge, such as an expert demonstration with annotations? The two application domains for this project are air traffic control, course scheduling and a real-time strategy game called Wargus.
End-user debugging of Machine Learning Systems with Margaret Burnett and Simone Stumpf
How do you debug a program that was written by a machine instead of a person, especially when you do not know much about programming or machine learning, and are working with a program you cannot even see? This is the problem faced by users of a new type of program being used today, namely, machine learning systems that, after being deployed, customize themselves by learning from an end-user's behavior. Prime examples of these programs include adaptive user interfaces, intelligent desktop assistants, email classifiers, and recommender systems. Inaccurate predictions by these learning systems erode users' trust and curb widespread acceptance of such systems. We seek to improve both the performance and acceptance of machine learning systems by allowing end-users to debug these learned programs when they make incorrect predictions.

Mining sequential and structured data

Automated Enterprise Event Monitoring
IT staff are finding it increasingly difficult to maintain the IT infrastructures due to its growing complexity. Many IT components are now instrumented with monitors that raise alerts when certain events occur. These alerts typically end up at an operator's console, but the sheer volume of alerts results in many of them being ignored. We are investigating ways to use machine learning to help network engineers deal with these alerts.
Species distribution mapping with Tom Dietterich, Matthew Betts and Julia Jones
There are many factors that influence the presence of species in a geographic location. These factors include environmental factors (eg. land cover, vegetation composition), climate, presence/absence of other species, spatial structure of the landscape, and life history traits of the species. We are developing models based on machine learning that incorporate these various factors when producing species distribution maps.
Bioinformatics
I am currently working with Todd Mockler (Botany and Plant Pathology) and Xiaoli Fern on discovering genetic regulatory networks that control circadian and diurnal patterns in plants. In addition, I am collaborating with Andy Karplus (BioPhysics) on learning accurate models of protein geometry.
Disease Outbreak Detection with Greg Cooper (University of Pittsburgh)
I've been involved in the field of biosurveillance since graduate school. The goal of disease outbreak detection is to use health-care data from a city to detect as quickly and as accurately as possible the onset of a disease outbreak. I don't do too much work in this area anymore.

Conference Deadlines

AAAI 2010 (Atlanta, GA): Abstracts - Jan 18, 2010, Full Submission - Jan 21, 2010
KDD 2010 (Washington, D.C): Abstracts - Feb 2, 2010, Full Submission - Feb 5, 2010
ICML 2010 (Haifa, Israel): Full Submission - Feb 1, 2010