Events · OBD 2018
Optimization and Big Data 2018
KAUST Research Workshop · 5–7 February 2018 · Thuwal, Saudi Arabia
Conference Center Hall, Building 19, Level 3
Organizers: Peter Richtárik and Marco Canini
Keynote: Tamas Terlaky (Lehigh University)
Original KAUST page · Wayback copy
Fourth workshop in the series, on optimization algorithms and distributed systems that scale to problems with billions of variables, using distributed and parallel computing, randomization, asynchronicity, decomposition, sketching, and streaming. Funded by the KAUST Office of Sponsored Research, co-sponsored by the Alan Turing Institute, with additional support from the KAUST Industry Collaboration Program.
Speakers
- Tamas Terlaky (Lehigh) — 60 Years of Interior Point Methods: From Periphery to Glory
- Guillaume Obozinski (École des Ponts - ParisTech) — An SDCA-Powered Inexact Dual Augmented Lagrangian Method for Fast CRF Learning
- Martin Jaggi (EPFL) — Learning in a Distributed and Heterogeneous Environment
- Peter Richtárik (KAUST) — Stochastic Reformulations of Linear and Convex Feasibility Problems: Algorithms and Convergence Theory
- Olivier Fercoq (Télécom ParisTech) — Convergence Speed of a Primal-Dual Coordinate Descent Method
- Tamas Terlaky (Lehigh) — A Polynomial-time Rescaled von Neumann Algorithm for Linear Feasibility Problems
- Katya Scheinberg (Lehigh) — Direct and Efficient Optimization of Prediction Error and AUC of Linear Classifiers
- Bernard Ghanem (KAUST) — FFTLasso: Large-Scale Lasso in the Fourier Domain
- Jared Tanner (Oxford) — Sparse Non-Negative Super-Resolution: Simplified and Stabilized
- Wolfgang Heidrich (KAUST) — Optimization and Big Data in Computational Imaging
- Aritra Dutta (KAUST) — Online and Batch Supervised Background Estimation via L1 Regression
- Raphael Hauser (Oxford) — Emitter Array Tomosynthesis via Nonlinear Compressed Sensing
- Martin Takáč (Lehigh) — SARAH: A Novel Method for Machine Learning Problems Using Stochastic Recursive Gradient
- Taesoo Kim (Georgia Tech) — Processing a Trillion-Edge Graph on a Single Machine
- Marco Canini (KAUST) — In-Network Computation is a Dumb Idea Whose Time Has Come
- Panos Kalnis (KAUST) — Pivoted Subgraph Isomorphism – The Optimist, the Pessimist and the Realist
- Srikanth Kandula (Microsoft Research) — Approximate Answers for Complex Parallel Queries
- Matthias Ehrhardt (Cambridge) — Stochastic PDHG with Arbitrary Sampling and Applications to Medical Imaging
- George Lan (Georgia Tech) — Communication-Efficient Methods for Decentralized and Stochastic Optimization
- Ion Necoara (Politehnica University of Bucharest) — Conditions for Linear Convergence of (Stochastic) First Order Methods
Program
5 February 2018
- 09:15–09:30 Welcome and opening remarks
- 09:30–10:30 Tamas Terlaky
- 11:00–11:30 Guillaume Obozinski
- 11:30–12:00 Martin Jaggi
- 13:45–14:15 Peter Richtárik
- 14:15–16:00 Spotlight talks and poster session
- 16:00–16:30 Olivier Fercoq
- 16:30–17:00 Tamas Terlaky
6 February 2018
- 09:30–10:00 Katya Scheinberg
- 10:00–10:30 Bernard Ghanem
- 11:00–11:30 Jared Tanner
- 11:30–12:00 Wolfgang Heidrich
- 13:45–14:15 Aritra Dutta
- 14:15–16:00 Spotlight talks and poster session
- 16:00–16:30 Raphael Hauser
- 16:30–17:00 Martin Takáč
7 February 2018
- 09:30–10:00 Taesoo Kim
- 10:00–10:30 Marco Canini
- 11:00–11:30 Panos Kalnis
- 11:30–12:00 Srikanth Kandula
- 13:45–14:15 Matthias Ehrhardt
- 14:15–16:00 Spotlight talks and poster session
- 16:00–16:30 George Lan
- 16:30–17:00 Ion Necoara
- 18:00–18:15 Best poster prize ceremony
Best poster: Steffen Maass (Georgia Tech), Mosaic: Processing a Trillion-Edge Graph on a Single Machine.
Best KAUST poster: Adel Bibi (KAUST), High Order Tensor Formulation for Convolutional Sparse Coding.