The Melodi Laboratory at the University of Washington, Seattle (Department of Electrical and Computer Engineering, or ECE) comprises a group of individuals studying machine learning, optimization, and data interpretation. We look at machine learning, artificial intelligence, and data science with twin goals: (1) advance the theoretical and mathematical understanding of new and existing algorithms to understand why they work; and (2) conducting empirically-driven data science experiments on ever increasingly large real-world data sets, and working on system-level strategies to achieve this. We are interested in discrete and continuous mathematical optimization and statistical strategies for solving some of the most challenging AI problems today. We work on a number of applications, including speech (SP) and natural language processing (NLP), computational biology, computer vision, and sensor networks.
UW Melodi-Lab Founder
Jeff Bilmes, Professor, University of Washington, Seattle, Dept. of ECE, Dept. of CSE (adjunct), and Dept. of Linguistics (adjunct)
Research
Artificial Intelligence
In our view, AI simply means that computing systems perform complex tasks on real-world signals effectively, and move beyond what traditionally programmed computers have done.
Machine Learning
Machine learning has so far proven itself to be the best way to achieve artificial intelligence like capabilities. Since humans are unable to directly program such complex tasks, we can instead employ mathematical optimization to produce programs in the form of learnt models.
Submodular/Supermodular Analysis
Much of machine learning involves continuous optimization (e.g., stochastic gradient descent), but many problems are inherently discrete, such as feature and data subset selection, structure learning, summarization, coresets, sketching, sparse models and so on. Submodularity and supermodularity is a natural and fast paradigm to study hard discrete optimization problems and apply them to real-world tasks.
Speech and Language Processing
Speech and language are amazing natural signals that humans easily produce and consume. How can we get computers to do the same?
Graphical Models
Graphical models are a formal, mathematical, but visually concise and intuitive way to visualize important properties of families of probability distributions. They also facilitate the development of practical and scalable strategies to compute probabilistic inference overs such families for many machine learning problems.
Data Science
Scientists at the Melodi lab work on other problems in computational biology (genomics, proteomics), computer vision, sensor networks, network computing, smart cities, the science of data management, computational music, high-performance computing, and so on. See our publications list for more.
