Publications

. Statistical Guarantees for Local Graph Clustering. arXiv, 2019.

Preprint

. Parallel and Communication Avoiding Least Angle Regression. arXiv, 2019.

Preprint

. Locality And Structure Aware Graph Node Embedding. WI2018, 2018.

Preprint PDF

. Variational Perspective on Local Graph Clustering. Math. Program., 2017.

Preprint PDF Code Slides Video

. Capacity Releasing Diffusion for Speed and Locality. ICML, 2017.

Preprint PDF Code Video

. Avoiding communication in primal and dual block coordinate descent methods. SIAM SISC, 2017.

Preprint PDF Slides Video

. A Randomized Rounding Algorithm for Sparse PCA. TKDD, 2017.

Preprint PDF

. Parallel Local Graph Clustering. VLDB., 2016.

Preprint PDF Code Slides

. Matrix-free interior point method for compressed sensing problems. Math. Program., 2014.

Preprint PDF Code Slides

Code

  • Local Graph Clustering provides methods to find local clusters in a given graph without touching the whole graph. C++ implementation with Python interface.

  • Trillion. Instance generators for l1-regularized over- and underdetermined least squares. The generators are implemented in MATLAB, they are memoryless and computationally inexpensive. Hence, large-scale instances can be created.

  • pdNCG: primal-dual Newton Conjugate Gradients. A MATLAB implementation for the solution of unconstrained l1-regularized problems. For example, Machine Learning problems, such as l1-regularized least-squares and logistic regression, Compressed Sensing problems, such as l1-synthesis, l1-analysis and isotropic total-variation. The solver is memoryless, it requires only matrix-vector product operations, hence it is appropriate for large-scale instances.

  • MFIPMCS: Matrix-free Interior Point Method for Compressed Sensing. An interior point method implemented in MATLAB for the solution of real valued compressed sensing problems. The “matrix-free” implies that only matrix-vector products operations are allowed and the process is memoryless. The solver implements efficient preconditioning techniques for the fast solution of linear systems at every iteration.

  • FCD: Flexible Coordinate Descent. This software only reproduces the experiments in paper Flexible Coordinate Descent.

Recent & Upcoming Talks

Local Graph Clustering
Jul 14, 2019 2:00 PM
A Short Introduction to Software for Local Graph Clustering
Jun 14, 2019 2:00 PM
Variational Perspective on Local Graph Clustering
Jul 6, 2018 6:00 PM
Numerical Optimization, Formulations, Serial and Parallel Algorithms for Machine Learning and Network Science
Jun 20, 2018 3:00 PM
Parallel Methods for Convex Optimization
Dec 14, 2017 10:00 AM
Communication Avoiding Optimization Algorithms
May 22, 2017 12:00 AM
Parallel Local Graph Clustering
Sep 21, 2016 4:00 PM
Local graph clustering and optimization
Jun 21, 2016 3:00 PM
Performance of First- and Second-Order Methods for Large-Scale Optimization
Jul 12, 2015 2:45 PM
Primal-Dual Newton Conjugate Gradients Method for Compressed Sensing Problems
Mar 16, 2015 12:00 AM
Newton-type Methods for Big-Data Optimization
Feb 25, 2015 12:00 AM
Robust Block Coordinate Descent
Dec 16, 2014 12:00 AM
Robust Block Coordinate Descent
Dec 1, 2014 12:00 AM
A Second-Order Method for Strongly Convex L1-regularisation Problems
Sep 4, 2014 12:00 AM
A Second-Order Method for Sparse Signal Reconstruction in Compressed Sensing
Jul 13, 2014 12:00 AM
Primal-Dual Newton Conjugate Gradients Method for L1-regularized Problems
Jul 12, 2014 12:00 AM
A Second-Order Method for Sparse Signal Reconstruction in Compressed Sensing
May 19, 2014 12:00 AM
Second-Order Methods for l1-Regularized Problems in Machine Learning and Sparse Signal Reconstruction
Oct 31, 2013 12:00 AM
Second-Order Methods for Signal Reconstruction, Big-Data and Machine Learning Problems
Jul 30, 2013 12:00 AM
A Second-Order Method for Strongly Convex l1-Regularization Problems
Jul 3, 2013 12:00 AM
A Second-Order Method for Strongly Convex l1-Regularization Problems
Jul 2, 2013 12:00 AM
Second-Order Methods for Sparse Signal Reconstruction
Jun 21, 2013 12:00 AM
Preconditioning Linear Systems in Interior Point Methods for Sparse Signal Reconstruction
Sep 11, 2012 12:00 AM
Matrix-free Interior Point Method for Compressed Sensing Problems
Aug 19, 2012 12:00 AM
Matrix-free Interior Point Method for Compressed Sensing Problems
May 13, 2012 12:00 AM

Projects

Numerical Optimization, Formulations and Algorithms, for Machine Learning

NSERC-Discovery Grant. Project duration 2019-2025.

Robust, Efficient, And Localizable Machine Learning Primitives

Data-Driven Discovery of Models (D3M). Project duration 2016-2020. In collaboration with M. Mahoney, Farbod Roosta-Khorasani and Alex Gittens.

News

  • (06/10/2019) I will be a panelist of a Round Table Discussion about clustering algorithms at PASC 2019. This is part of the minisymposium “Identifying Relevant Communities in Immense Networks: Clustering Algorithms that Leverage High-Performance Computing”.
  • (03/22/2019) Our session on Optimization Methods for Graph Analytics at ICCOPT with Di Wang has been accepted. Speakers will be Shai Ben-David, Rasmus Kyng, Nate Veldt, Anastasios Zouzias, Di Wang and myself.
  • (02/21/2019) I will give a talk on our Local Graph Clustering API at PASC 2019. The title of the minisymposium is “Identifying Relevant Communities in Immense Networks: Clustering Algorithms that Leverage High-Performance Computing”.
  • (11/30/2018) Our paper Locality And Structure Aware Graph Node Embedding was awarded the Best Student Paper Award at WI2018.
  • (08/09/2018) I will be a member of the Technical Program Committee at SC19.

Students

I am always looking for highly motivated and hard-working Master’s and PhD students to work with. Send me an email if you would like to work with me.

  • Chufeng Hu, MS, 2019 -

Teaching

Course: CS 370 Winter 2019 - Numerical Computation

Contact