Home/Andrew Moore

Andrew Moore

Professor and Dean
Email: awm@andrew.cmu.edu
Office: GHC 5113
Phone: (412) 334-2063
Personal Homepage: http://www.cs.cmu.edu/~awm

My research focuses entirely on applying machine learning to control. I want to find out just how autonomous we can make robots, factories and other complex control systems.

A vast range of complex systems are becoming amenable to computer decision making. Most such systems have sensors, and so through their lifetime they experience a stream of sensor data. The machine analysis of this data stream can lead to the system improving its own internal knowledge of its behavior. And in turn it can use the improved knowledge to design improved control laws for itself. How can we do this mining of the data stream to extract as much predictive accuracy as possible, how can we do it quickly, and how can we then make the system develop good control laws with what it has learned?

I work with these questions by developing a number of experimental algorithms, described shortly, and be testing them with a somewhat large variety of real-world problems. The large variety is to find out whether the algorithms are sufficiently autonomous that I can get away without being an expert in the application domains. Some completed, ongoing, and future applications include a billiards robot, robot juggling, candy bar manufacture, RC helicopters, decision making in economics, textile manufacture, RC toy cars, pinball, electricity pricing, car engine emission control and computer games.

The kind of algorithms I work with lie at the borders between artificial intelligence, statistics, control theory and heuristic search. I make heavy use of a class of techniques called memory-based algorithms which have some useful properties. Predictions are made with interpolations from local previous experiences, and can be performed very quickly using geometric techniques, can circumvent the problems of a one-to-many inverse in redundant systems and avoid the necessity of retraining behaviors which have not been used recently.

Another key to fast learning is exploration. Both the learning effort and the collection of data should be concentrated on the critical parts of the problem. Furthermore, real-world tasks can impose severe demands on learning controllers. I am developing the general memory-based learning (GMBL) system to tackle many of these difficulties. This technique centers on efficiently harnessing an amazingly powerful, but normally hopelessly expensive, statistical technique called leave-one-out cross-validation. This can help the systems automatically select the most appropriate function approximator, the best trade-off between fitting the data and filtering noise, and can be made to detect irrelevant, coupled or partially relevant sensors and actuators. Inventing good techniques to perform large-scale cross-validation searches is an exciting area with ongoing research opportunities.

To generate improved control laws autonomously, the system must be able to make long-term predictions, and plan to achieve long-term goals. This planning must happen whilst simultaneously learning the environment. Weak-but-general search methods such as Dynamic Programming and A* search can prove helpful here, while retaining wide applicability to many domains. Some recent research involves scheduling search control during on-line planning to minimize wasted computations, and there are further interesting possibilities there. Ongoing research extends this to the PartiGame algorithm, in which exploration and planning are scheduled in a multi-resolution manner. A recursive partitioning of state-space adapts itself in real time to yield fine detail in the critical regions, while remaining at a coarse granularity elsewhere.

I am also interested in working on several new developments of the research described above, ranging from (1) developing a quickly re-configurable meta-robot system with which we can produce, test and learn new control tasks with a turnover period of days instead of months, (2) developing formal theories of classes of control tasks, with associated proofs of convergence of appropriate machine learning algorithms, (3) developing new learning algorithms for certain classes of harder problems, and (4) tailoring efficient versions of certain optimization techniques, including genetic algorithms, to the control domain.

Displaying 79 Publications
Fast State Discovery for HMM Model Selection and Learning
Sajid Siddiqi, Geoffrey Gordon and Andrew Moore

Conference Paper, Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics (AI-STATS), January, 2007
Scalable and robust group discovery on large transactional data
Pak Yan Choi, Andrew Moore and Jeremy Martin Kubica

Tech. Report, CMU-RI-TR-05-60, Robotics Institute, Carnegie Mellon University, December, 2005
Variable KD-Tree Algorithms for Spatial Pattern Search and Discovery
Jeremy Martin Kubica, Joseph Masiero, Andrew Moore, Robert Jedicke and Andrew J. Connolly

Conference Paper, Carnegie Mellon University, Neural Information Processing Systems, December, 2005
Variable KD-Tree Algorithms for Efficient Spatial Pattern Search
Jeremy Martin Kubica, Joseph Masiero, Andrew Moore, Robert Jedicke and Andrew J. Connolly

Tech. Report, CMU-RI-TR-05-43, Robotics Institute, Carnegie Mellon University, September, 2005
A Multiple Tree Algorithm for the Efficient Association of Asteroid Observations
Jeremy Martin Kubica, Andrew Moore, Andrew J. Connolly and Robert Jedicke

Conference Paper, Carnegie Mellon University, The Eleventh ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 138-146, August, 2005
Efficiently Identifying Close Track/Observation Pairs in Continuous Timed Data
Jeremy Martin Kubica, Andrew Moore, Andrew J. Connolly and Robert Jedicke

Conference Paper, Carnegie Mellon University, Proc. SPIE Signal and Data Processing of Small Targets, August, 2005
Fast Inference and Learning in Large-State-Space HMMs
Sajid Siddiqi and Andrew Moore

Conference Paper, Carnegie Mellon University, Proceedings of the 22nd International Conference on Machine Learning (ICML), August, 2005
Alias Detection in Link Data Sets
Paul Hsiung, Andrew Moore, Daniel Neill and Jeff Schneider

Conference Paper, Carnegie Mellon University, Proceedings of the International Conference on Intelligence Analysis, May, 2005
Making Logistic Regression A Core Data Mining Tool: A Practical Investigation of Accuracy, Speed, and Simplicity
Paul Komarek and Andrew Moore

Tech. Report, CMU-RI-TR-05-27, Robotics Institute, Carnegie Mellon University, May, 2005
Efficient Algorithms for the Identification of Potential Track/Observation Associations in Continuous Time Data
Jeremy Martin Kubica, Andrew Moore, Andrew J. Connolly and Robert Jedicke

Tech. Report, CMU-RI-TR-05-10, Robotics Institute, Carnegie Mellon University, February, 2005
Fast and Robust Track Initiation Using Multiple Trees
Jeremy Martin Kubica, Andrew Moore, Andrew J. Connolly and Robert Jedicke

Tech. Report, CMU-RI-TR-04-62, Robotics Institute, Carnegie Mellon University, November, 2004
Spatial Data Structures for Efficient Trajectory-Based Queries
Jeremy Martin Kubica, Andrew Moore, Andrew J. Connolly and Robert Jedicke

Tech. Report, CMU-RI-TR-04-61, Robotics Institute, Carnegie Mellon University, November, 2004
Fast Nonlinear Regression via Eigenimages Applied to Galactic Morphology
Brigham Anderson, Andrew Moore, Andrew J. Connolly and Robert Nichol

Conference Paper, Carnegie Mellon University, International Conference on Knowledge Discovery and Data Mining, August, 2004
Semantic based Biomedical Image Indexing and Retrieval
Yanxi Liu, Nicole Lazar, W.E. Rothfus, Frank Dellaert, Andrew Moore, Jeff Schneider and Takeo Kanade

Book Section/Chapter, Carnegie Mellon University, Trends and Advances in Content-Based Image and Video Retrieval, February, 2004
Probabilistic Noise Identification and Data Cleaning
Jeremy Martin Kubica and Andrew Moore

Conference Paper, Carnegie Mellon University, The Third IEEE International Conference on Data Mining, pp. 131-138, November, 2003
Tractable Group Detection on Large Link Data Sets
Jeremy Martin Kubica, Andrew Moore and Jeff Schneider

Conference Paper, Carnegie Mellon University, The Third IEEE International Conference on Data Mining, pp. 573-576, November, 2003
K-groups: Tractable Group Detection on Large Link Data Sets
Jeremy Martin Kubica, Andrew Moore and Jeff Schneider

Tech. Report, CMU-RI-TR-03-32, Robotics Institute, Carnegie Mellon University, September, 2003
cGraph: A Fast Graph-Based Method for Link Analysis and Queries
Jeremy Martin Kubica, Andrew Moore, David Cohn and Jeff Schneider

Conference Paper, Carnegie Mellon University, Proceedings of the 2003 IJCAI Text-Mining & Link-Analysis Workshop, pp. 22-31, August, 2003
Finding Underlying Connections: A Fast Graph-Based Method for Link Analysis and Collaboration Queries
Jeremy Martin Kubica, Andrew Moore, David Cohn and Jeff Schneider

Conference Paper, Carnegie Mellon University, Proceedings of the 2003 International Conference on Machine Learning, pp. 392-399, August, 2003
Probabilistic Noise Identification and Data Cleaning
Jeremy Martin Kubica and Andrew Moore

Tech. Report, CMU-RI-TR-02-26, Robotics Institute, Carnegie Mellon University, October, 2002
Stochastic Link and Group Detection
Jeremy Martin Kubica, Andrew Moore, Jeff Schneider and Yiming Yang

Conference Paper, Carnegie Mellon University, The Eighteenth National Conference on Artificial Intelligence, pp. 798-804, August, 2002
Classification-Driven Pathological Neuroimage Retrieval Using Statistical Asymmetry Measures
Yanxi Liu, Frank Dellaert, William E. Rothfus, Andrew Moore, Jeff Schneider and Takeo Kanade

Conference Paper, Carnegie Mellon University, Proceedings of the 2001 Medical Imaging Computing and Computer Assisted Intervention Conference (MICCAI '01), October, 2001
Mix-nets: Factored Mixtures of Gaussians in Bayesian Networks with Mixed Continuous and Discrete Variables
Scott Davies and Andrew Moore

Tech. Report, CMU-CS-00-119, Computer Science Department, Carnegie Mellon University, April, 2000
Q2: memory-based active learning for optimizing noisy continuous functions
Andrew Moore, Jeff Schneider, Justin Boyan and M.S. Lee

Conference Paper, Carnegie Mellon University, IEEE International Conference on Robotics and Automation (ICRA '00), Vol. 4, pp. 4095 - 4102, April, 2000
A Locally Weighted Learning Tutorial using Vizier 1.0
Jeff Schneider and Andrew Moore

Tech. Report, CMU-RI-TR-00-18, Robotics Institute, Carnegie Mellon University, February, 2000
The Anchors Hierarchy: Using the triangle inequality to survive high dimensional data
Andrew Moore

Tech. Report, CMU-RI-TR-00-05, Robotics Institute, Carnegie Mellon University, February, 2000
Influence and Variance of a Markov Chain: Application to Adaptive Discretization in Optimal Control
Remi Munos and Andrew Moore

Conference Paper, IEEE Conference on Decision and Control, Vol. 2, pp. 1464 - 1469, December, 1999
Variable resolution discretization for high-accuracy solutions of optimal control problems
Remi Munos and Andrew Moore

Conference Paper, International Joint Conference on Artificial Intelligence, August, 1999
Cached Sufficient Statistics for Automated Mining and Discovery from Massive Data Sources
Andrew Moore, Jeff Schneider, Brigham Anderson, Scott Davies, Paul Komarek, Mary Soon Lee, Marina Meila, Remi Munos, Kary Myers and Dan Pelleg

Miscellaneous, July, 1999
Gradient Descent Approaches to Neural-Net-Based Solutions of the Hamilton-Jacobi-Bellman Equation
Remi Munos, Leemon Baird and Andrew Moore

Conference Paper, International Joint Conference on Neural Networks, July, 1999
Accelerating Exact k-means Algorithms with Geometric Reasoning
Dan Pelleg and Andrew Moore

Conference Paper, Knowledge Discovery from Databases (KDD '99), January, 1999
Bayesian Networks for Lossless Dataset Compression
Scott Davies and Andrew Moore

Conference Paper, 1999 Knowledge Discovery from Databases (KDD '99), January, 1999
Distributed Value Functions
Jeff Schneider, Weng-Keen Wong, Andrew Moore and Martin Riedmiller

Conference Paper, International Conference on Machine Learning, January, 1999
Efficient Multi-Object Dynamic Query Histograms
Mark Derthick, James Harrison, Andrew Moore and Steven F. Roth

Conference Paper, Proceedings of the IEEE Information Visualization Conference (InfoVis '99), January, 1999
Gradient Descent for General Reinforcement Learning
Leemon Baird and Andrew Moore

Journal Article, Advances in Neural Information Processing Systems 11, January, 1999
Multi-Value-Functions: Efficient Automatic Action Hierarchies for Multiple Goal MDPs
Andrew Moore, Leemon Baird and Leslie Pack Kaelbling

Conference Paper, Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI '99), January, 1999
Variable Resolution Discretization in Optimal Control
Remi Munos and Andrew Moore

Journal Article, Machine Learning Journal, January, 1999
Barycentric Interpolator for Continuous Space and Time Reinforcement Learning
Remi Munos and Andrew Moore

Conference Paper, Neural Information Processing Systems, Vol. 11, December, 1998
Stochastic production scheduling to meet demand forecasts
Jeff Schneider, Justin Boyan and Andrew Moore

Conference Paper, Proceedings of the 37th IEEE Conference on Decision and Control, Vol. 3, pp. 2722 - 2727, December, 1998
Very Fast EM-based Mixture Model Clustering using Multiresolution kd-trees
Andrew Moore

Journal Article, Neural Information Systems Processing, December, 1998
ADtrees for Fast Counting and for Fast Learning of Association Rules
Brigham Anderson and Andrew Moore

Journal Article, Knowledge Discovery from Databases '98, August, 1998
On the Greediness of Feature Selection Algorithms
Kan Deng and Andrew Moore

Conference Paper, International Conference of Machine Learning (ICML '98), June, 1998
Q2: Memory-based active learning for optimizing noisy continuous functions
Andrew Moore, Jeff Schneider, Justin Boyan and Mary Lee

Conference Paper, International Conference of Machine Learning, June, 1998
Cached Sufficient Statistics for Efficient Machine Learning with Large Datasets
Andrew Moore and Mary Soon Lee

Journal Article, Journal of Artificial Intelligence Research, Vol. 8, pp. 67- 91, March, 1998
Value Function Based Production Scheduling
Jeff Schneider, Justin Boyan and Andrew Moore

Conference Paper, Machine Learning: Proceedings of the Fifteenth International Conference (ICML '98), March, 1998
On Greediness of Feature Selection Algorithms
Kan Deng and Andrew Moore

Tech. Report, CMU-RI-TR-98-03, Robotics Institute, Carnegie Mellon University, February, 1998
Applying Online Search Techniques to Reinforcement Learning
Scott Davies, A. Y. Ng and Andrew Moore

Conference Paper, Fifteenth National Conference on Artificial Intelligence (AAAI), January, 1998
Learning Evaluation Functions for Global Optimization and Boolean Satisfiability
Justin Boyan and Andrew Moore

Conference Paper, Fifteenth National Conference on Artificial Intelligence, January, 1998
On-line Memory-based Detection of General Purpose Systems
Kan Deng, Andrew Moore and Michael Nechyba

Conference Paper, Neural Information Systems Processing 1998 (NIPS '98), January, 1998
Cached Sufficient Statistics for Efficient Machine Learning with Large Databases
Andrew Moore and Mary Soon Lee

Tech. Report, CMU-RI-TR-97-27, Robotics Institute, Carnegie Mellon University, July, 1997
Data Mining at CALD-CMU: Tools, Experiences and Research Directions.
C Faloutsos, G. Gibson, Tom Mitchell, Andrew Moore and Sebastian Thrun

Conference Paper, Proceedings of the AFCEA International's First Federal Data Mining Symposium, January, 1997
Efficient Locally Weighted Polynomial Regression Predictions
Andrew Moore, Jeff Schneider and Kan Deng

Conference Paper, International Conference on Machine Learning, January, 1997
Learning to Recognize Time Series: Combining ARMA models with Memory-based Learning
Kan Deng, Andrew Moore and Michael Nechyba

Conference Paper, IEEE Int. Symp. on Computational Intelligence in Robotics and Automation, Vol. 1, pp. 246 - 250, January, 1997
Locally Weighted Learning
C. G. Atkeson, S. A. Schaal and Andrew Moore

Journal Article, AI Review, Vol. 11, January, 1997
Locally Weighted Learning For Control
Andrew Moore, C. G. Atkeson and S. A. Schaal

Journal Article, AI Review, Vol. 11, pp. 75-113, January, 1997
Using Prediction to Improve Combinatorial Optimization Search
Justin Boyan and Andrew Moore

Conference Paper, Sixth International Workshop on Artificial Intelligence and Statistics, January, 1997
Learning Evaluation Functions for Large Acyclic Domains
Justin Boyan and Andrew Moore

Conference Paper, Machine Learning: Proceedings of the Thirteenth International Conference, January, 1996
Reinforcement Learning: A Survey
L.P. Kaelbling, M.L. Littman and Andrew Moore

Journal Article, Journal of Artificial Intelligence Research, Vol. 4, pp. 237-285, January, 1996
The Parti-game Algorithm for Variable Resolution Reinforcement Learning in Multidimensional State-spaces
Andrew Moore and C. G. Atkeson

Journal Article, Machine Learning, Vol. 21, December, 1995
Robust Value Function Approximation by Working Backwards
Justin Boyan and Andrew Moore

Conference Paper, Proceedings of the Workshop on Value Function Approximation, Machine Learning Conference, July, 1995
Learning Automated Product Recommendations Without Observable Features: An Initial Investigation
Mary S. Lee and Andrew Moore

Tech. Report, CMU-RI-TR-95-17, Robotics Institute, Carnegie Mellon University, April, 1995
Memory-Based Learning for Control
Andrew Moore, C. G. Atkeson and S. A. Schaal

Tech. Report, CMU-RI-TR-95-18, Robotics Institute, Carnegie Mellon University, April, 1995
Variable Resolution Reinforcement Learning
Andrew Moore

Tech. Report, CMU-RI-TR-95-19, Robotics Institute, Carnegie Mellon University, April, 1995
Generalization in Reinforcement Learning: Safely Approximating the Value Function
Justin Boyan and Andrew Moore

Conference Paper, Advances in Neural Information Processing Systems 7, January, 1995
Locally Weighted Bayesian Regression
Andrew Moore

Miscellaneous, January, 1995
Memory-based Stochastic Optimization
Andrew Moore and Jeff Schneider

Conference Paper, Neural Information Processing Systems 8, January, 1995
Multiresolution Instance-Based Learning
Andrew Moore, Jeff Schneider and Kan Deng

Conference Paper, Proceedings of International Joint Conference on Artificial Intelligence, January, 1995
An Empirical Investigation of Brute Force to choose Features, Smoothers and Function Approximators
Andrew Moore, D. J. Hill and M. P . Johnson

Journal Article, Computational Learning Theory and Natural Learning Systems, Vol. 3, January, 1994
Efficient Algorithms for Minimizing Cross Validation Error
Andrew Moore and M. S. Lee

Conference Paper, Proceedings of the 11th International Conference on Machine Learning, January, 1994
Prioritized Sweeping: Reinforcement Learning with Less Data and Less Real Time
Andrew Moore and C. G. Atkeson

Journal Article, Machine Learning, Vol. 13, October, 1993
Hoeffding Races: Accelerating Model Selection Search for Classification and Function Approximation
O. Maron and Andrew Moore

Conference Paper, Advances in Neural Information ProcessingSystems 6, January, 1993
Memory-based Reinforcement Learning: Efficient Computation with Prioritized Sweeping
Andrew Moore and C. G. Atkeson

Conference Paper, Advances in Neural Information Processing Systems 5, January, 1992
Knowledge of Knowledge and Intelligent Experimentation for Learning Control
Andrew Moore

Conference Paper, Proceedings of the 1991 Seattle International Joint Conference on Neural Networks, Vol. 2, pp. 683 - 688, July, 1991
Efficient Memory-based Learning for Robot Control
Andrew Moore

PhD Thesis, Robotics Institute, Carnegie Mellon University, No. Technical Report 209, March, 1991
An introductory tutorial on kd-trees
Andrew Moore

Tech. Report, Technical Report No. 209, Computer Laboratory, University of Cambridge, January, 1991
Fast, Robust Adaptive Control by Learning only Forward Models
Andrew Moore

Conference Paper, January, 1991
Variable Resolution Dynamic Programming: Efficiently Learning Action Maps in Multivariate Real-valued State-spaces
Andrew Moore

Conference Paper, Proceedings of the Eighth International Conference on Machine Learning, January, 1991
Acquisition of Dynamic Control Knowledge for a Robotic Manipulator
Andrew Moore

Conference Paper, Proceedings of the 7th International Conference on Machine Learning, January, 1990
Some Experiments in Adaptive State Space Robotics
W. F. Clocksin and Andrew Moore

Conference Paper, Proceedings of the 7th AISB Conference, January, 1989
Mailing Address
Carnegie Mellon University
Robotics Institute
5000 Forbes Avenue
Pittsburgh, PA 15213
2017-09-13T10:47:32+00:00