Michael Mahoney's new home page

Basic info I am at ICSI and the Department of Statistics at UC Berkeley, and I am also in the RISELab (in the past AMPLab) in the Department of Computer Science. You can reach me via electronic mail: mmahoney is the username, and stat dot berkeley dot edu follows the at symbol. (Of course, my gmail address is still valid.) Also, some other (somewhat outdated) information can be found at my old web page at Stanford and my older web page at Yale.

Students, postdocs, etc.

I have been fortunate enough to work with a number of great students and postdocs over the years.

Research interests

• Algorithmic and statistical aspects of modern large-scale data analysis.
• Randomized linear algebra and randomized numerical linear algebra.
• Scalable convex and non-convex optimization in parallel/distributed environments.
• Implicit regularization and implicit optimization methods in scalable approximation algorithms.
• Graph approximation algorithms and applications to large social and information networks.
• Applications to DNA microarray, SNP, astronomical, medical imaging, and other scientific data.

A lot of my work has focused on Randomized Linear Algebra, i.e., using random sampling and random projection methods to solve very large matrix-based problems; developing geometric network analysis tools, i.e., using scalable approximation algorithms with a geometric or statistical flavor to analyze the structure and dynamics of large informatics graphs; developing approximate computation and regularization methods for large informatics graphs; applications to community detection, clustering, and information dynamics in large social and information networks; and applications to DNA single nucleotide polymorphism (SNP) data, astronomical and medical imaging data, and large-scale statistical data analysis more generally.

In the past, I developed and analyzed algorithms for large matrix, graph, and regression problems, and I applied these and related tools to the statistical data analysis of extremely large scientific and Internet data sets. For example, I worked on large-scale web analytics, machine learning, and query log analysis; applications of graph partitioning algorithms to clustering and community identification; and applications of randomized matrix algorithms to hyperspectral medical image data, DNA microarray data, and DNA SNP data.

In the more distant past, I have also worked on developing and analyzing Monte Carlo algorithms for performing useful computations on extremely large matrices, e.g., the additive-error and relative-error CUR matrix decompositions. Past research has also included work in computational statistical mechanics on the development and analysis of the TIP5P model of liquid water, as well as work in both computational and experimental biophysics on proteins and protein-nucleic acid interactions.

Other items of interest

Software. (See publication list for code to reproduce results on any one paper.)

Teaching.

Workshops.

Publications

• Alchemist: An Apache Spark <=> MPI Interface,
A. Gittens, K. Rothauge, M. W. Mahoney, S. Wang, L. Gerhardt, Prabhat, J. Kottalam, M. Ringenburg, and K. Maschhoff,
Technical Report, Preprint: arXiv:1806.01270 (2018) (arXiv),
Accepted for publication, CUG 2018.
• Accelerating Large-Scale Data Analysis by Offloading to High-Performance Computing Libraries using Alchemist,
A. Gittens, K. Rothauge, S. Wang, M. W. Mahoney, L. Gerhardt, Prabhat, J. Kottalam, M. Ringenburg, and K. Maschhoff,
Technical Report, Preprint: arXiv:1805.11800 (2018) (arXiv),
Accepted for publication, KDD 2018.
• Group Collaborative Representation for Image Set Classification,
B. Liu, L. Jing, J. Li, J. Yu, A. Gittens, and M. W. Mahoney,
Accepted for publication, International Journal of Computer Vision.
• Error Estimation for Randomized Least-Squares Algorithms via the Bootstrap,
M. E. Lopes, S. Wang, M. W. Mahoney,
Technical Report, Preprint: arXiv:1803.08021 (2018) (arXiv),
Accepted for publication, ICML 2018,
Journal version submitted for publication.
• GPU Accelerated Sub-Sampled Newton's Method,
S. B. Kylasa, F. Roosta-Khorasani, M. W. Mahoney, and A. Grama,
Technical Report, Preprint: arXiv:1802.09113 (2018) (arXiv), (code),
Submitted for publication.
• Hessian-based Analysis of Large Batch Training and Robustness to Adversaries,
Z. Yao, A. Gholami, Q. Lei, K. Keutzer, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1802.08241 (2018) (arXiv),
Submitted for publication.
• Inexact Non-Convex Newton-Type Methods,
Z. Yao, P. Xu, F. Roosta-Khorasani, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1802.06925 (2018) (arXiv),
Submitted for publication.
• Out-of-sample extension of graph adjacency spectral embedding,
K. Levin, F. Roosta-Khorasani, M. W. Mahoney, and C. E. Priebe,
Technical Report, Preprint: arXiv:1802.06307 (2018) (arXiv),
Accepted for publication, ICML 2018,
Journal version submitted for publication.

• Lectures on Randomized Numerical Linear Algebra,
P. Drineas and M. W. Mahoney,
Technical Report, Preprint: arXiv:1712.08880 (2017) (arXiv),
To appear in: Lectures of the 2016 PCMI Summer School on Mathematics of Data.
• Avoiding Synchronization in First-Order Methods for Sparse Convex Optimization,
A. Devarakonda, K. Fountoulakis, J. Demmel, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1712.06047 (2017) (arXiv),
Accepted for publication, Proc. IPDPS 2018.
• Rethinking generalization requires revisiting old ideas: statistical mechanics approaches and complex learning behavior, (click here for a blog about this paper)
C. H. Martin and M. W. Mahoney,
Technical Report, Preprint: arXiv:1710.09553 (2017) (arXiv),
• LASAGNE: Locality And Structure Aware Graph Node Embedding,
E. Faerman, F. Borutta, K. Fountoulakis, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1710.06520 (2017) (arXiv),
• A Berkeley View of Systems Challenges for AI,
I. Stoica, D. Song, R. A. Popa, D. A. Patterson, M. W. Mahoney, R. H. Katz, A. D. Joseph, M. Jordan, J. M. Hellerstein, J. Gonzalez, K. Goldberg, A. Ghodsi, D. E. Culler, and P. Abbeel,
Technical Report No. UCB/EECS-2017-159, October 2017 (www),
• GIANT: Globally Improved Approximate Newton Method for Distributed Optimization,
S. Wang, F. Roosta-Khorasani, P. Xu, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1709.03528 (2017) (arXiv), (Spark code), (Python code),
Submitted for publication.
• Second-Order Optimization for Non-Convex Machine Learning: An Empirical Study,
P. Xu, F. Roosta-Khorasani, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1708.07827 (2017) (arXiv), (code),
Submitted for publication.
• Newton-Type Methods for Non-Convex Optimization Under Inexact Hessian Information,
P. Xu, F. Roosta-Khorasani, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1708.07164 (2017) (arXiv),
Submitted for publication.
• A Bootstrap Method for Error Estimation in Randomized Matrix Multiplication,
M. E. Lopes, S. Wang, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1708.01945 (2017) (arXiv),
Submitted for publication.
• Capacity releasing diffusions for speed and locality,
D. Wang, K. Fountoulakis, M. Henzinger, M. W. Mahoney, and S. Rao,
Technical Report, Preprint: arXiv:1706.05826 (2017) (arXiv),
Proc. of the 34th ICML Conference 3598-3607 (2017) (pdf), (supp).
• Scalable Kernel K-Means Clustering with Nystrom Approximation: Relative-Error Bounds,
S. Wang, A. Gittens, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1706.02803 (2017) (arXiv),
Accepted for publication, J. Machine Learning Research.
• Union of Intersections (UoI) for Interpretable Data Driven Discovery and Prediction,
K. E. Bouchard, A. F. Bujan, F. Roosta-Khorasani, S. Ubaru, Prabhat, A. M. Snijders, J.-H. Mao, E. F. Chang, M. W. Mahoney, S. Bhattacharyya,
Technical Report, Preprint: arXiv:1705.07585 (2017) (arXiv),
Proc. of the 2017 NIPS Conference (pdf).
• Skip-Gram - Zipf + Uniform = Vector Additivity,
A. Gittens, D. Achlioptas, and M. W. Mahoney,
Proc. of the 55th ACL Meeting 69-76 (2017) (pdf).
• Principles and Applications of Science of Information [Scanning the Issue],
T. Courtade, A. Grama, M. W. Mahoney, and T. Weissman,
Proceedings of the IEEE, 105(2): 183-188 (2017) (pdf).
• Social Discrete Choice Models,
D. Zhang, K. Fountoulakis, J. Cao, M. Yin, M. W. Mahoney, and A. Pozdnoukhov,
Technical Report, Preprint: arXiv:1703.07520 (2017) (arXiv),
Submitted for publication.
• Sketched Ridge Regression: Optimization Perspective, Statistical Perspective, and Model Averaging,
S. Wang, A. Gittens, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1702.04837 (2017) (arXiv),
Proc. of the 34th ICML Conference 3608-3616 (2017) (pdf),
J. Machine Learning Research, 18(218): 1-50 (2018) (pdf).

• Avoiding communication in primal and dual block coordinate descent methods,
A. Devarakonda, K. Fountoulakis, J. Demmel, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1612.04003 (2016) (arXiv),
Submitted for publication.
• Feature-distributed sparse regression: a screen-and-clean approach,
J. Yang, M. W. Mahoney, M. A. Saunders, and Y. Sun,
Proc. of the 2016 NIPS Conference (pdf).
• Multi-label learning with semantic embeddings,
L. Jing, M. Cheng, L. Yang, A. Gittens, M. W. Mahoney,
ICLR 2017 OpenReview.net (www),
• Mapping the Similarities of Spectra: Global and Locally-biased Approaches to SDSS Galaxy Data,
D. Lawlor, T. Budavari, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1609.03932 (2016) (arXiv),
The Astrophysical Journal, 833:1, 26 (2016) (pdf).
• Lecture Notes on Spectral Graph Methods,
M. W. Mahoney,
Technical Report, Preprint: arXiv:1608.04845 (2016) (arXiv),
• Lecture Notes on Randomized Linear Algebra,
M. W. Mahoney,
Technical Report, Preprint: arXiv:1608.04481 (2016) (arXiv),
• An optimization approach to locally-biased graph algorithms,
K. Fountoulakis, D. F. Gleich, M. W. Mahoney,
Technical Report, Preprint: arXiv:1607.04940 (2016) (arXiv),
Proceedings of the IEEE, 105(2): 256-272 (2017) (pdf).
• DCAR: A Discriminative and Compact Audio Representation to Improve Event Detection,
L. Jing, B. Liu, J. Choi, A. Janin, J. Bernd, M. W. Mahoney, and G. Friedland,
Technical Report, Preprint: arXiv:1607.04378 (2016) (arXiv),
Proc. of the 2016 ACM Multimedia Conference 57-61 (2016) (pdf),
IEEE Transactions on Multimedia, 19(12): 2637-2650 (2017) (pdf).
• Matrix Factorization at Scale: a Comparison of Scientific Data Analytics in Spark and C+MPI Using Three Case Studies,
A. Gittens, A. Devarakonda, E. Racah, M. Ringenburg, L. Gerhardt, J. Kottalam, J. Liu, K. Maschhoff, S. Canon, J. Chhugani, P. Sharma, J. Yang, J. Demmel, J. Harrell, V. Krishnamurthy, M. W. Mahoney, and Prabhat,
Technical Report, Preprint: arXiv:1607.01335 (2016) (arXiv), (code),
Proc. 2016 IEEE BigData, 204-213 (2016) (pdf).
• Sub-sampled Newton Methods with Non-uniform Sampling,
P. Xu, J. Yang, F. Roosta-Khorasani, C. Re, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1607.00559 (2016) (arXiv),
Proc. of the 2016 NIPS Conference (pdf).
• Approximating the Solution to Mixed Packing and Covering LPs in parallel time,
M. W. Mahoney, S. Rao, D. Wang, and P. Zhang,
Proc. of the 43rd ICALP Conference, 52:1-52:14 (2016) (pdf).
• A Simple and Strongly-Local Flow-Based Method for Cut Improvement,
N. Veldt, D. F. Gleich, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1605.08490 (2016) (arXiv),
Proc. of the 33rd ICML Conference 1938-1947 (2016) (pdf), (supp).
• RandNLA: Randomized Numerical Linear Algebra,
P. Drineas and M. W. Mahoney,
Communications of the ACM, 59, 80-90 (2016) (pdf).
• FLAG n' FLARE: Fast Linearly-Coupled Adaptive Gradient Methods,
X. Cheng, F. Roosta-Khorasani, S. Palombo, P. L. Bartlett, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1605.08108 (2016) (arXiv),
Proc. of the 21st International Conference on AISTATS, PMLR 84:404-414 (2018) (pdf, supp).
• Parallel Local Graph Clustering,
J. Shun, F. Roosta-Khorasani, K. Fountoulakis, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1604.07515 (2016) (arXiv),
Proceedings of the VLDB Endowment, 9(12) 1041-1052 (2016) (pdf).
• A multi-platform evaluation of the randomized CX low-rank matrix factorization in Spark,
A. Gittens, J. Kottalam, J. Yang, M. F. Ringenburg, J. Chhugani, E. Racah, M. Singh, Y. Yao, C. Fischer, O. Ruebel, B. Bowen, N. G. Lewis, M. W. Mahoney, V. Krishnamurthy, and Prabhat,
Proc. 5th International Workshop on Parallel and Distributed Computing for Large Scale Machine Learning and Big Data Analytics, at IPDPS, 2016 (pdf).
• Mining Large Graphs,
D. F. Gleich and M. W. Mahoney,
In Handbook of Big Data. pp. 191-220, edited by P. Buhlmann, P. Drineas, M. Kane, and M. van de Laan, Chapman and Hall/CRC Press, 2016 (pdf).
• Structural properties underlying high-quality Randomized Numerical Linear Algebra algorithms,
M. W. Mahoney and P. Drineas,
In Handbook of Big Data. pp. 137-154, edited by P. Buhlmann, P. Drineas, M. Kane, and M. van de Laan, Chapman and Hall/CRC Press, 2016 (pdf).
• Variational Perspective on Local Graph Clustering,
K. Fountoulakis, X. Cheng, J. Shun, F. Roosta-Khorasani and M. W. Mahoney,
Technical Report, Preprint: arXiv:1602.01886 (2016) (arXiv),
Accepted for publication, Mathematical Programming.
• Sub-Sampled Newton Methods II: Local Convergence Rates,
F. Roosta-Khorasani and M. W. Mahoney,
Technical Report, Preprint: arXiv:1601.04738 (2016) (arXiv),
Journal version submitted for publication.
• Sub-Sampled Newton Methods I: Globally Convergent Algorithms,
F. Roosta-Khorasani and M. W. Mahoney,
Technical Report, Preprint: arXiv:1601.04737 (2016) (arXiv),
Journal version submitted for publication.
• RandNLA, Pythons, and the CUR for Your Data Problems: Reporting from G2S3 2015 in Delphi,
E. Gallopoulos, P. Drineas, I. Ipsen, and M. W. Mahoney,
SIAM News 49:1 January/February 2016 (web), (pdf).

• Faster Parallel Solver for Positive Linear Programs via Dynamically-Bucketed Selective Coordinate Descent,
D. Wang, M. W. Mahoney, N. Mohan, and S. Rao,
Technical Report, Preprint: arXiv:1511.06468 (2015) (arXiv).
• A Local Perspective on Community Structure in Multilayer Networks,
L. G. S. Jeub, M. W. Mahoney, P. J. Mucha, and M. A. Porter,
Technical Report, Preprint: arXiv:1510.05185 (2015) (arXiv),
Network Science, 5(2): 144-163, 2017 (pdf).
• Optimal Subsampling Approaches for Large Sample Linear Regression,
R. Zhu, P. Ma, M. W. Mahoney, and B. Yu,
Technical Report, Preprint: arXiv:1509.05111 (2015) (arXiv).
• Unified Acceleration Method for Packing and Covering Problems via Diameter Reduction,
D. Wang, S. Rao, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1508.02439 (2015) (arXiv),
Proc. of the 43rd ICALP Conference, 50:1-50:13 (2016) (pdf).
• Using local spectral methods to robustify graph-based learning algorithms,
D. F. Gleich and M. W. Mahoney,
Proc. of the 21st Annual SIGKDD, (2015) (pdf) (code).
• Structured Block Basis Factorization for Scalable Kernel Matrix Evaluation,
R. Wang, Y. Li, M. W. Mahoney, and E. Darve,
Technical Report, Preprint: arXiv:1502.03571 (2015) (arXiv).
• Identifying important ions and positions in mass spectrometry imaging data using CUR matrix decompositions,
J. Yang, O. Rubel, Prabhat, M. W. Mahoney, and B. P. Bowen,
Analytical Chemistry, 87 (9), 4658-4666 (2015) (pdf) (code).
• Spectral Gap Error Bounds for Improving CUR Matrix Decomposition and the Nystrom Method,
D. G. Anderson, S. S. Du, M. W. Mahoney, C. Melgaard, K. Wu, and M. Gu,
Proc. of the 18th International Conference on AISTATS, PMLR 38:19-27 (2015) (pdf, supp) (code).
• Weighted SGD for Lp Regression with Randomized Preconditioning,
J. Yang, Y.-L. Chow, C. Re, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1502.03571 (2015) (arXiv),
Proc. of the 27-th Annual SODA, 558-569 (2016) (pdf),
J. Machine Learning Research, 18(211): 1-43 (2018) (pdf).
• Implementing Randomized Matrix Algorithms in Parallel and Distributed Environments,
J. Yang, X. Meng, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1502.03032 (2015) (arXiv) (code),
Proceedings of the IEEE 104(1): 58-92 (2016) (pdf).

• Tree decompositions and social graphs,
A. B. Adcock, B. D. Sullivan, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1411.1546 (2014) (arXiv), (code).
Internet Mathematics, 12(5), 315-361 (2016) (pdf).
• Fast Randomized Kernel Methods With Statistical Guarantees,
A. El Alaoui and M. W. Mahoney,
Technical Report, Preprint: arXiv:1411.0306 (2014) (arXiv),
Proc. of the 2015 NIPS Conference (pdf).
• Signal Processing for Big Data (Editorial for Special Issue)
G. B. Giannakis, F. Bach, R. Cendrillon, M. Mahoney, and J. Neville,
IEEE Signal Processing Magazine, 31: 15-16 (September 2014) (pdf).
• A Statistical Perspective on Randomized Sketching for Ordinary Least-Squares,
G. Raskutti and M. W. Mahoney,
Technical Report, Preprint: arXiv:1406.5986 (2014) (arXiv),
Proc. of the 32nd ICML Conference (2015) (pdf),
J. Machine Learning Research, 17(214): 1-31, 2016 (pdf).
• Random Laplace Feature Maps for Semigroup Kernels on Histograms,
J. Yang, V. Sindhwani, Q. Fan, H. Avron, and M. W. Mahoney,
Proc. of the 27th CVPR Conference, 971-978 (2014) (pdf).
• Anti-differentiating Approximation Algorithms: A case study with Min-cuts, Spectral, and Flow,
D. F. Gleich and M. W. Mahoney,
Proc. of the 31st ICML Conference, JMLR W&CP 32 (1): 1018-1025 (2014) (pdf) (code, code) (talk).
• Quasi-Monte Carlo Feature Maps for Shift-Invariant Kernels,
J. Yang, V. Sindhwani, H. Avron, and M. W. Mahoney,
Proc. of the 31st ICML Conference, JMLR W&CP 32 (1): 485-493 (2014) (pdf), (code),
Technical Report, Preprint: arXiv:1412.8293 (2014) (arXiv),
J. Machine Learning Research, 17(120): 1-38 (2016) (pdf).
• Think Locally, Act Locally: The Detection of Small, Medium-Sized, and Large Communities in Large Networks,
L. G. S. Jeub, P. Balachandran, M. A. Porter, P. J. Mucha, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1403.3795 (2014) (arXiv), (code, code),
Physical Review E, 91, 012821 (2015) (pdf).
• A new spin on an old algorithm: technical perspective on "Communication costs of Strassen's matrix multiplication,"
M. W. Mahoney,
Communications of the ACM, 57(2): 106 (2014) (pdf).

• Tree-like Structure in Large Social and Information Networks,
A. B. Adcock, B. D. Sullivan, and M. W. Mahoney,
Proc. of the 2013 IEEE ICDM, 1-10 (2013) (pdf).
• Objective Identification of Informative Wavelength Regions in Galaxy Spectra,
C.-W. Yip, M. W. Mahoney, A. S. Szalay, I. Csabai, T. Budavari, R. F. G. Wyse, and L. Dobos,
Technical Report, Preprint: arXiv:1312.0637 (2013) (arXiv),
Astronomical Journal, 147, 5, 110 (2014) (pdf).
• Evaluating OpenMP Tasking at Scale for the Computation of Graph Hyperbolicity,
A. B. Adcock, B. D. Sullivan, O. R. Hernandez, and M. W. Mahoney,
Proc. of the 9-th IWOMP, 71-83 (2013) (pdf).
• Frontiers in Massive Data Analysis,
Committee on the Analysis of Massive Data, et al. (M. I. Jordan, et al.),
The National Academies Press (2013) (pdf), (web).
• A Statistical Perspective on Algorithmic Leveraging,
P. Ma, M. W. Mahoney, and B. Yu,
Technical Report, Preprint: arXiv:1306.5362 (2013) (arXiv),
Proc. of the 31st ICML Conference, JMLR W&CP 32 (1): 91-99 (2014) (pdf),
J. Machine Learning Research, 16, 861-911 (2015) (pdf).
• Robust Regression on MapReduce,
X. Meng, and M. W. Mahoney,
Proc. of the 30th ICML Conference, JMLR W&CP 28(3): 888-896 (2013) (pdf).
• Quantile Regression for Large-scale Applications,
J. Yang, X. Meng, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1305.0087 (2013) (arXiv), (code),
Proc. of the 30th ICML Conference, JMLR W&CP 28(3): 881-887 (2013) (pdf),
SIAM J. Scientific Computing, 36(5), S78-S110 (2014) (pdf).
• Revisiting the Nystrom Method for Improved Large-Scale Machine Learning,
A. Gittens and M. W. Mahoney,
Technical Report, Preprint: arXiv:1303.1849 (2013) (arXiv), (code),
Proc. of the 30th ICML Conference, JMLR W&CP 28(3): 567-575 (2013) (pdf),
J. Machine Learning Research, 17(117): 1-65 (2016) (pdf).

• Semi-supervised Eigenvectors for Large-scale Locally-biased Learning,
T. J. Hansen and M. W. Mahoney,
Proc. of the 2012 NIPS Conference (pdf), (code),
Technical Report, Preprint: arXiv:1304.7528 (2013) (arXiv),
J. Machine Learning Research, 15, 3691-3734 (2014) (pdf).
• Low-distortion Subspace Embeddings in Input-sparsity Time and Applications to Robust Linear Regression,
X. Meng and M. W. Mahoney,
Technical Report, Preprint: arXiv:1210.3135 (2012) (arXiv),
Proc. of the 45-th STOC, 91-100 (2013) (pdf).
• The Fast Cauchy Transform and Faster Robust Linear Regression,
K. L. Clarkson, P. Drineas, M. Magdon-Ismail, M. W. Mahoney, X. Meng, and D. P. Woodruff,
Technical Report, Preprint: arXiv:1207.4684 (2012) (arXiv),
Proc. of the 24-th Annual SODA, 466-477 (2013) (pdf),
SIAM J. Computing, 45, 763-810 (2016) (pdf).
• rCUR: an R package for CUR matrix decomposition,
A. Bodor, I. Csabai, M. W. Mahoney, and N. Solymosi,
BMC Bioinformatics, 13:103 (2012) (pdf), (code).
• Approximate Computation and Implicit Regularization for Very Large-scale Data Analysis,
M. W. Mahoney,
Technical Report, Preprint: arXiv:1203.0786 (2012) (arXiv),
Proc. of the 2012 ACM Symposium on Principles of Database Systems, 143-154, 2012 (pdf).
• On the Hyperbolicity of Small-World and Tree-Like Random Graphs,
W. Chen, W. Fang, G. Hu, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1201.1717 (2012) (arXiv),
Proc. of the 23-rd ISAAC 278-288 (2012) (pdf),
Internet Mathematics, 9(4), 434-491 (2013) (pdf).

• Randomized Dimensionality Reduction for K-means Clustering,
C. Boutsidis, A. Zouzias, M. W. Mahoney, and P. Drineas,
Technical Report, Preprint: arXiv:1110.2897 (2011) (arXiv),
IEEE Transactions on Information Theory, 61(2), 1045-1062 (2015) (pdf).
• Regularized Laplacian Estimation and Fast Eigenvector Approximation,
P. O. Perry and M. W. Mahoney,
Technical Report, Preprint: arXiv:1110.1757 (2011) (arXiv),
Proc. of the 2011 NIPS Conference (pdf).
• LSRN: A Parallel Iterative Solver for Strongly Over- or Under-Determined Systems,
X. Meng, M. A. Saunders, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1109.5981 (2011) (arXiv), (code),
SIAM J. Scientific Computing, 36(2), C95-C118 (2014) (pdf).
• Fast approximation of matrix coherence and statistical leverage,
P. Drineas, M. Magdon-Ismail, M. W. Mahoney, and D. P. Woodruff,
Technical Report, Preprint: arXiv:1109.3843 (2011) (arXiv),
Proc. of the 29th ICML Conference (2012) (pdf),
J. Machine Learning Research, 13, 3475-3506 (2012) (pdf).
• Localization on low-order eigenvectors of data matrices,
M. Cucuringu and M. W. Mahoney,
Technical Report, Preprint: arXiv:1109.1355 (2011) (arXiv).
• Efficient Genomewide Selection of PCA-Correlated tSNPs for Genotype Imputation,
A. Javed, P. Drineas, M. W. Mahoney, and P. Paschou,
Annals of Human Genetics, 75, 707-722 (2011) (pdf).
• Randomized Algorithms for Matrices and Data,
M. W. Mahoney,
Foundations and Trends in Machine Learning, NOW Publishers, Volume 3, Issue 2, 2011 (now),
TR version: Technical Report, Preprint: arXiv:1104.5557 (2011) (arXiv).
(Abridged version in: Advances in Machine Learning and Data Mining for Astronomy, edited by M. J. Way, et al., pp. 647-672, 2012.)

• Computation in Large-Scale Scientific and Internet Data Applications is a Focus of MMDS 2010,
M. W. Mahoney,
Technical Report, Preprint: arXiv:1012.4231 (2010) (arXiv),
Appeared in SIGKDD Explorations, SIGACT News, ASA-SCGN Newsletter, and IMS Bulletin.
• CUR from a Sparse Optimization Viewpoint,
J. Bien, Y. Xu, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1011.0413 (2010) (arXiv),
Proc. of the 2010 NIPS Conference (ps, pdf).
• Algorithmic and Statistical Perspectives on Large-Scale Data Analysis,
M. W. Mahoney,
Technical Report, Preprint: arXiv:1010.1609 (2010) (arXiv),
In: Combinatorial Scientific Computing, pp. 427-469, edited by U. Naumann and O. Schenk, 2012.
• Implementing regularization implicitly via approximate eigenvector computation,
M. W. Mahoney and L. Orecchia,
Technical Report, Preprint: arXiv:1010.0703 (2010) (arXiv),
Proc. of the 28th ICML Conference, 121-128 (2011) (pdf) (talk).
• Approximating Higher-Order Distances Using Random Projections,
P. Li, M. W. Mahoney, and Y. She,
Proc. of the 26th UAI Conference, 312-321 (2010) (ps, pdf),
Technical Report, Preprint: arXiv:1203.3492 (2012) (arXiv).
• Effective Resistances, Statistical Leverage, and Applications to Linear Equation Solving,
P. Drineas and M. W. Mahoney,
Technical Report, Preprint: arXiv:1005.3097 (2010) (arXiv).
• Empirical Comparison of Algorithms for Network Community Detection,
J. Leskovec, K. J. Lang, and M. W. Mahoney,
Technical Report, Preprint: arXiv:1004.3539 (2010) (arXiv),
Proc. of the 19-th International WWW, 631-640 (2010) (ps, pdf).

• A Local Spectral Method for Graphs: with Applications to Improving Graph Partitions and Exploring Data Graphs Locally,
M. W. Mahoney, L. Orecchia, and N. K. Vishnoi,
Technical Report, Preprint: arXiv:0912.0681 (2009) (arXiv),
J. Machine Learning Research, 13, 2339-2365 (2012) (ps, pdf).
• Unsupervised Feature Selection for the k-means Clustering Problem,
C. Boutsidis, M. W. Mahoney, and P. Drineas,
Proc. of the 2009 NIPS Conference (ps, pdf).
• Learning with Spectral Kernels and Heavy-Tailed Data,
M. W. Mahoney and H. Narayanan,
Technical Report, Preprint: arXiv:0906.4539 (2009) (arXiv).
• Empirical Evaluation of Graph Partitioning Using Spectral Embeddings and Flow,
K. J. Lang, M. W. Mahoney, and L. Orecchia,
Proc. of the 8-th International SEA, 197-208 (2009) (ps, pdf).
• CUR Matrix Decompositions for Improved Data Analysis,
M. W. Mahoney and P. Drineas,
Proc. Natl. Acad. Sci. USA, 106, 697-702 (2009) (ps, pdf).

• An Improved Approximation Algorithm for the Column Subset Selection Problem,
C. Boutsidis, M. W. Mahoney, and P. Drineas,
Technical Report, Preprint: arXiv:0812.4293 (2008) (arXiv),
Proc. of the 20-th Annual SODA, 968-977 (2009) (ps, pdf).
• Algorithmic and Statistical Challenges in Modern Large-Scale Data Analysis are the Focus of MMDS 2008
M. W. Mahoney, L.-H. Lim, and G. E. Carlsson
Technical Report, Preprint: arXiv:0812.3702 (2008) (arXiv),
Appeared in SIGKDD Explorations (ps, pdf), SIAM News (ps, pdf), and ASA-SCGN Newsletter (ps, pdf), and abridged versions appeared in IMS Bulletin (ps, pdf) and AmStat News.
• Community Structure in Large Networks: Natural Cluster Sizes and the Absence of Large Well-Defined Clusters,
J. Leskovec, K. J. Lang, A. Dasgupta, and M. W. Mahoney,
Technical Report, Preprint: arXiv:0810.1355 (2008) (arXiv),
Internet Mathematics, 6(1), 29-123 (2009) (pdf).
• Unsupervised Feature Selection for Principal Components Analysis,
C. Boutsidis, M. W. Mahoney, and P. Drineas,
Proc. of the 14-th Annual SIGKDD, 61-69 (2008) (ps, pdf).
• Statistical Properties of Community Structure in Large Social and Information Networks,
J. Leskovec, K. J. Lang, A. Dasgupta, and M. W. Mahoney,
Proc. of the 17-th International WWW, 695-704 (2008) (ps, pdf).

• Faster Least Squares Approximation,
P. Drineas, M. W. Mahoney, S. Muthukrishnan, and T. Sarlos,
Technical Report, Preprint: arXiv:0710.1435 (2007) (arXiv),
Numerische Mathematik, 117, 219-249 (2011) (pdf).
• PCA-Correlated SNPs for Structure Identification in Worldwide Human Populations,
P. Paschou, E. Ziv, E. G. Burchard, S. Choudhry, W. Rodriguez-Cintron, M. W. Mahoney, and P. Drineas,
PLoS Genetics, 3, 1672-1686 (2007) (ps, pdf).
• Relative-Error CUR Matrix Decompositions,
P. Drineas, M. W. Mahoney, and S. Muthukrishnan,
Technical Report, Preprint: arXiv:0708.3696 (2007) (arXiv),
SIAM J. Matrix Analysis and Applications, 30, 844-881 (2008) (ps, pdf).
• Feature Selection Methods for Text Classification,
A. Dasgupta, P. Drineas, B. Harb, V. Josifovski, and M. W. Mahoney,
Proc. of the 13-th Annual SIGKDD, 230-239 (2007) (ps, pdf).
• Sampling Algorithms and Coresets for Lp Regression,
A. Dasgupta, P. Drineas, B. Harb, R. Kumar, and M. W. Mahoney,
Technical Report, Preprint: arXiv:0707.1714 (2007) (arXiv),
Proc. of the 19-th Annual SODA, 932-941 (2008) (ps, pdf),
SIAM J. Computing, 38, 2060-2078 (2009) (ps, pdf).
• Web Information Retrieval and Linear Algebra Algorithms,
A. Frommer, M. W. Mahoney, and D. B. Szyld (Eds.),
Proc. of Dagstuhl Seminar 07071, (2007) (web).
• Intra- and interpopulation genotype reconstruction from tagging SNPs,
P. Paschou, M. W. Mahoney, A. Javed, J. R. Kidd, A. J. Pakstis, S. Gu, K. K. Kidd, and P. Drineas,
Genome Research, 17(1), 96-107 (2007) (ps, pdf).

• Bridging the Gap Between Numerical Linear Algebra, Theoretical Computer Science, and Data Applications,
G. H. Golub, M. W. Mahoney, P. Drineas, and L.-H. Lim,
SIAM News 39:8 October 2006 (ps, pdf).
• Randomized Algorithms for Matrices and Massive Data Sets,
P. Drineas and M. W. Mahoney,
Proc. of the 32-nd Annual VLDB, 1269 (2006) (ps, pdf).
• Subspace Sampling and Relative-Error Matrix Approximation: Column-Row-Based Methods,
P. Drineas, M. W. Mahoney, and S. Muthukrishnan,
Proc. of the 14-th Annual ESA, 304-314 (2006) (ps, pdf).
• Subspace Sampling and Relative-Error Matrix Approximation: Column-Based Methods,
P. Drineas, M. W. Mahoney, and S. Muthukrishnan,
Proc. of the 10-th Annual RANDOM, 316-326 (2006) (ps, pdf).
• Tensor-CUR Decompositions For Tensor-Based Data,
M. W. Mahoney, M. Maggioni, and P. Drineas,
Proc. of the 12-th Annual SIGKDD, 327-336 (2006) (ps, pdf),
SIAM J. Matrix Analysis and Applications, 30, 957-987 (2008) (ps, pdf).
• Polynomial Time Algorithm for Column-Row-Based Relative-Error Low-Rank Matrix Approximation,
P. Drineas, M. W. Mahoney, and S. Muthukrishnan,
Technical Report, DIMACS TR 2006-04 March 2006 (ps, pdf).
• Sampling Algorithms for L2 Regression and Applications,
P. Drineas, M. W. Mahoney, and S. Muthukrishnan,
Proc. of the 17-th Annual SODA, 1127-1136 (2006) (ps, pdf).

• A Randomized Algorithm for a Tensor-Based Generalization of the Singular Value Decomposition,
P. Drineas and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1327, June 2005 (ps, pdf),
Linear Algebra and its Applications, 420, 553-571 (2007) (ps, pdf).
• On the Nystrom Method for Approximating a Gram Matrix for Improved Kernel-Based Learning,
P. Drineas and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1319, April 2005 (ps, pdf),
Proc. of the 18-th Annual COLT, 323-337 (2005) (ps, pdf),
J. Machine Learning Research, 6, 2153-2175 (2005) (ps, pdf).

• Sampling Sub-problems of Heterogeneous Max-Cut Problems and Approximation Algorithms,
P. Drineas, R. Kannan, and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1283, April 2004 (ps, pdf),
Proc. of the 22-nd Annual STACS, 57-68 (2005) (ps, pdf),
Random Structures and Algorithms, 32:3, 307-333 (2008) (ps, pdf).
• Fast Monte Carlo Algorithms for Matrices III: Computing an Efficient Approximate Decomposition of a Matrix,
P. Drineas, R. Kannan, and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1271, February 2004 (ps, pdf),
SIAM J. Computing, 36, 184-206 (2006) (ps, pdf).
• Fast Monte Carlo Algorithms for Matrices II: Computing Low-Rank Approximations to a Matrix,
P. Drineas, R. Kannan, and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1270, February 2004 (ps, pdf),
SIAM J. Computing, 36, 158-183 (2006) (ps, pdf).
• Fast Monte Carlo Algorithms for Matrices I: Approximating Matrix Multiplication,
P. Drineas, R. Kannan, and M. W. Mahoney,
Technical Report, YALEU/DCS/TR-1269, February 2004 (ps, pdf),
SIAM J. Computing, 36, 132-157 (2006) (ps, pdf).

• Rapid Mixing of Several Markov Chains for a Hard-Core Model,
R. Kannan, M. W. Mahoney, and R. Montenegro,
Proc. of the 14-th Annual ISAAC, 663-675 (2003) (pdf).

• Quantum, Intramolecular Flexibility, and Polarizability Effects on the Reproduction of the Density Anomaly of Liquid Water by Simple Potential Functions,
M. W. Mahoney and W. L. Jorgensen,
J. Chem. Phys., 115, 10758-10768 (2001) (pdf).
• Rapid Estimation of Electronic Degrees of Freedom in Monte Carlo Calculations for Polarizable Models of Liquid Water,
M. W. Mahoney and W. L. Jorgensen,
J. Chem. Phys., 114, 9337-9349 (2001) (pdf).
• Diffusion Constant of the TIP5P Model of Liquid Water,
M. W. Mahoney and W. L. Jorgensen,
J. Chem. Phys., 114, 363-366 (2001) (pdf).

• A Five-Site Model for Liquid Water and the Reproduction of the Density Anomaly by Rigid, Nonpolarizable Potential Functions,
M. W. Mahoney and W. L. Jorgensen,
J. Chem. Phys., 112, 8910-8922 (2000) (pdf).

• Repression and Activation of Promoter-Bound RNA Polymerase Activity by Gal Repressor,
H. E. Choy, R. R. Hanger, T. Aki, M. Mahoney, K. Murakami, A. Ishihama, and S. Adhya,
J. Mol. Biol. 272: 293-300, 1997 (pdf).
• Discrete Representations of the Protein C-alpha Chain,
X. F. de la Cruz, M. W. Mahoney, and B. K. Lee,
Fold. & Des. 2: 223-234, 1997 (pdf).

MMDS Workshops

I run the MMDS meetings. We started the MMDS Workshops on "Algorithms for Modern Massive Data Sets" to address algorithmic and statistical challenges in modern large-scale statistical data analysis.

• MMDS 2016 took place on the campus of UC Berkeley on June 21-24, 2016. See the main MMDS web page for more information.
• MMDS 2014 took place on the campus of UC Berkeley on June 17-20, 2014. See the main MMDS web page for more information. Click here for the entire video collection.
• MMDS 2012 took place on the campus of Stanford University on July 10-13, 2012. For pdfs and videos of the presentations, go to the main MMDS web page; or click here for the entire video collection.
• MMDS 2010 took place on the campus of Stanford University on June 15-18, 2010. MMDS 2010 addressed computation in large-scale scientific and internet data applications more generally. See the MMDS web page for details, including articles and all the speaker overheads!
• MMDS 2008 took place on June 25-28, 2008. MMDS 2008 grew out of our expectation for what the algorithmic and statistical foundations of large-scale data analysis should look like a generation from now. Click here for an article that appeared in SIGKDD Explorations and SIAM News about the meeting.
• MMDS 2006 took place on June 21-24, 2006. MMDS 2006 was originally motivated by the complementary perspectives brought by numerical linear algebra and theoretical computer science to matrix algorithms in large-scale data applications. Click here for an article in SIAM News about the meeting.

Talks and presentations

Several recent talks:

Several tutorial presentations:

Several other older talks:

The TIP5P Water talk:

Videos of several talks:

My old web page at Yale has additional (somewhat outdated) information; and my less old web page at Stanford has additional (somewhat less outdated) information; but you should click on them (or Google search for them) if you are interested in extremely large homes.