Workshops

Fall 2017

Fast Iterative Methods in Optimization

Oct 2, 2017 to Oct 6, 2017

Return to event »

Click on the titles of individual talks for abstract, slides and archived video.

All events take place in the Calvin Lab auditorium.

Monday, October 2nd, 2017
9:00 am–9:20 am		Coffee and Check-In
9:20 am–9:30 am		Opening Remarks
9:30 am–10:15 am		Continuous Methods for Discrete Optimization: From Convex Relaxations, to Iterative Schemes, and Back Aleksander Mądry, MIT
10:15 am–10:45 am		Break
10:45 am–11:30 am		Width-independent Iterative Algorithms for Packing and Covering Programs Lorenzo Orecchia, Boston University
11:30 am–12:15 pm		Slime Molds and Sparse Recovery Nisheeth Vishnoi, École Polytechnique Fédérale de Lausanne
12:15 pm–2:00 pm		Lunch
2:00 pm–2:45 pm		Michael Cohen and ell_p Regression Yin-Tat Lee, University of Washington
2:45 pm–3:15 pm		Break
3:15 pm–4:00 pm		Chordal Graphs and Sparse Semidefinite Optimization Lieven Vandenberghe, UCLA
4:00 pm–4:45 pm		On the Optimization Landscape of Matrix and Tensor Decomposition Problems Tengyu Ma, Princeton University
4:45 pm–5:45 pm		Reception

Tuesday, October 3rd, 2017
9:00 am–9:30 am		Coffee and Check-In
9:30 am–10:15 am		The State of Contemporary Computing Substrates for Optimization Methods Ben Recht, UC Berkeley
10:15 am–10:45 am		Break
10:45 am–11:30 am		Let's Make Block Coordinate Descent Go Fast Mark Schmidt, University of British Columbia
11:30 am–12:15 pm		Stochastic, Second-order Black-box Optimization for Step Functions Katya Scheinberg, Lehigh University
12:15 pm–2:00 pm		Lunch
2:00 pm–2:45 pm		Algorithmic Tools for Smooth Nonconvex Optimization Steve Wright, University of Wisconsin-Madison
2:45 pm–3:15 pm		Break
3:15 pm–4:00 pm		Zero-order and Dynamic Sampling Methods for Nonlinear Optimization Jorge Nocedal, Northwestern University
4:00 pm–4:45 pm		Dealing with Linear Constraints via Random Permutation Ruoyu Sun, University of Illinois at Urbana-Champaign

Wednesday, October 4th, 2017
9:00 am–9:30 am		Coffee and Check-In
9:30 am–10:15 am		Sketching as a Tool for Fast Algorithm Design Alex Andoni, Columbia University
10:15 am–10:45 am		Break
10:45 am–11:30 am		Sublinear Time Low-rank Approximation of Positive Semidefinite Matrices David Woodruff, IBM Almaden
11:30 am–12:15 pm		Hashing-based-estimators for Kernel Density in High Dimensions Moses Charikar, Stanford University
12:15 pm–2:00 pm		Lunch
2:00 pm–2:45 pm		Randomized Iterative Methods and Complexity for Markov Decision Process Mengdi Wang, Princeton University
2:45 pm–3:15 pm		Break
3:15 pm–4:00 pm		Sketchy Decisions: Convex Low-Rank Matrix Optimization with Optimal Storage Madeleine Udell, Cornell University
4:00 pm–5:00 pm		Open Problems Session

Thursday, October 5th, 2017
9:00 am–9:30 am		Coffee and Check-In
9:30 am–10:15 am		Trends in Large-scale Nonconvex Optimization Suvrit Sra, MIT
10:15 am–10:45 am		Break
10:45 am–11:30 am		Faster Algorithms and New Iterative Methods for Computing the Stationary Distribution Aaron Sidford, Stanford University
11:30 am–12:15 pm		Nonconvex Optimization for High-dimensional Learning: From ReLUs to Submodular Maximization Mahdi Soltanolkotabi, University of Southern California
12:15 pm–2:00 pm		Lunch
2:00 pm–2:10 pm		Monte Carlo Approximation Certificates for K-means Clustering Soledad Villar, New York University
2:10 pm–2:20 pm		Approximate Gaussian Elimination for Laplacians Rasmus Kyng, Yale University
2:20 pm–2:30 pm		Michael Cohen's Proper Version of Nesterov's Accelerated Method Jelena Diakonikolas, Boston University
2:30 pm–2:40 pm		A Lyapunov Analysis for Accelerated Gauss-Seidel Stephen Tu, UC Berkeley
2:40 pm–2:50 pm		Optimization Problems over Submodular Polytopes Swati Gupta, MIT
2:50 pm–3:00 pm		An Iterative Quasi-polynomial Time Algorithm for Solving Parity Games Marcin Jurdzinski, University of Warwick
3:00 pm–3:30 pm		Break
3:30 pm–3:40 pm		Highly Efficient Algorithms for Lasso-type Problems Based on Second-order Sparsity Kim-Chuan Toh, National University of Singapore
3:40 pm–3:50 pm		The Douglas--Rachford Splitting Algorithm for Inconsistent Minimization Problems Walaa Moursi, University of British Columbia
3:50 pm–4:00 pm		On the Convergence and Complexity of Nonconvex ADMM Shiqian Ma, UC Davis
4:00 pm–4:10 pm		Faster Constrained Optimization via Approximate Projections Ludwig Schmidt, MIT

Friday, October 6th, 2017
9:00 am–9:30 am		Coffee and Check-In
9:30 am–10:15 am		Will Vanishing Gradients Ever Vanish from Deep Learning? Moritz Hardt, UC Berkeley
10:15 am–10:45 am		Break
10:45 am–11:30 am		From Minimum Cut to Submodular Minimization: Leveraging the Decomposable Structure Alina Ene, Boston University
11:30 am–12:15 pm		Natasha 2: Faster Non-convex Optimization Than SGD Zeyuan Allen-Zhu, Microsoft Research