Alex Gray
Alexander GRAY
Machine learning, statistics, and data mining
Computational mathematics for massive datasets
Applications: computational science and engineering
After completing Bachelor's degrees in Applied Mathematics (concentration in Computational Statistics) and Computer Science from UC Berkeley, spending summers at the Santa Fe Institute and Los Alamos National Laboratory, among other places, I worked for 6 years in the Machine Learning Systems Group of NASA's Jet Propulsion Laboratory. I completed my PhD on 4/29/03 in Computer Science after 3.6 years at Carnegie Mellon University, advised by Prof. Andrew Moore and I'm currently a Postdoctoral Fellow in the Robotics Institute, but starting in the fall I'll be an Assistant Professor at Georgia Tech.
Caffe Strada, Berkeley Los Alamos National Laboratory Mars rovers at the JPL Spacecraft Assembly Facility Newell-Simon Hall atrium, CMU
agray @
3128 Newell-Simon Hall
(412) 268-8014
C.V. [pdf] [ps]
My work focuses on developing the new statistical and computational principles demanded by next-generation challenges in data analysis and autonomy. I develop new methods directly driven by, informed by, and validated by hard real problems, often by bridging technical intuitions or mathematical concepts from distant fields and perspectives. Two of the main challenges of modern computational science which keep me awake at night are massive datasets and various curses of dimensionality. I have been concerned with computational strategies for dealing with the fundamental summations, integrals, and maximizations at the root of a wide variety of statistics and machine learning methods. My most recent research, in progress, is about alternative statistical theory for certain old problems. A long-term target is to nail down enough powerful and general statistical and computational foundational primitives to achieve automatic data analysis and data-driven control, a bottom-up path to fully autonomous systems.
Generalized N-body Methods
Fast kernel density estimation
Fast n-point correlation functions
Fast all-nearest-neighbors
Fast Gaussian process regression
Fast nonparametric Bayes classification
CDM simulation
Computational Astrophysics, Cosmology, Astronomy
Evidence for dark energy
The origin of galaxies
Quasar mapping
Large-scale structure of the universe
Galaxy morphologies and clusters
Computational Chemistry, Drug Discovery, Biology
Molecule ranking for virtual screening
Microarray data analysis
Two discrete random variate generators
Adaptive Monte Carlo Methods
Multi-tree Monte Carlo
High-dimensional integration without Markov chains
Proximity Search
The Proximity Project
Fast nearest-neighbor classification
Fast approximate nearest-neighbor
Tutorial: Data Structures for Fast Statistics
Voyager near moon of Jupiter
Advanced Systems
Autonomous planetary science by rover teams
Efficient computer systems
The changing constant
Derivation of Learning Algorithms
Automatic derivation of new EM algorithms
Cirque du Soleil 'O'
Other pursuits...