2018-03-28
Optimization, Statistics and Uncertainty
source: Simons Institute 2017年11月28日
This workshop aims to explore the emerging connections between optimization, statistics and machine learning. It will discuss ties, both well-known and anticipated, between optimization and various approaches to modeling, reasoning about and coping with uncertainty. Some of the key topics that will be covered are: unsupervised machine learning, computational hardness in statistics, differential privacy and its applications in statistics and machine learning, online regret minimization, robust optimization and planted models.
For more information please visit: https://simons.berkeley.edu/workshops...
These presentations were supported in part by an award from the Simons Foundation.
1 1:07:53 Data Driven Optimization Models and Algorithms
2 32:12 Regularized Nonlinear Acceleration
3 31:44 Learning One-hidden-layer Neural Networks with Landscape Design
4 33:32 Algorithmic Stability for Interactive Data Analysis
5 27:00 Preventing Overfitting in Adaptive Data Analysis via Stability
6 30:41 The Convergence of Hamiltonian Monte Carlo
7 33:49 Distributional Robustness, Learning, and Empirical Likelihood
8 58:02 An Instability in Variational Methods for Learning Topic Models
9 32:33 A Few Connections Between Optimization and Probability
10 28:46 Implicit Regularization in Nonconvex Statistical Estimation
11 36:37 Low-rank Matrix Completion: Adaptive Sampling Can Help When, How?
12 1:01:40 Graph Powering
13 30:28 On Approximation Guarantees for Greedy Low Rank Optimization
14 29:53 Fast Rates for Bandit Optimization with Upper-Confidence Frank-Wolfe
15 1:01:19 Optimization's Untold Gift to Learning: Implicit Regularization
16 32:33 The Power and Limitations of Kernel Learning
17 29:27 Sampling Algorithms for Combinatorial Pure Exploration
18 59:07 Compositional Properties of Statistical Procedures: An Information-Theoretic View
19 28:02 Stability and Convergence Trade-Off of Iterative Optimization Algorithms
20 30:14 Accuracy First: Selecting a Differential Privacy Level for Accuracy-Constrained...
Hierarchies, Extended Formulations and Matrix-Analytic Techniques
source: Simons Institute 2017年11月6日
An important development in the area of convex relaxations has been the introduction of systematic ways of strengthening them by lift-and-project techniques. This leads to several hierarchies of LP/SDP relaxations: Lovasz-Schrijver, Sherali-Adams and Sum of Squares (also known as the Lasserre hierarchy). The benefits and limitations of these hierarchies have been studied extensively over the last decade. Recently, strong negative results have been obtained, not only for specific hierarchies but even for the more general notion of extended formulations.
In this workshop we investigate the power and limitations of LP/SDP hierarchies as well as general extended formulations, and their ties to convex algebraic geometry. We also explore tools and concepts from matrix analysis with strong connections to SDP formulations: matrix concentration, matrix multiplicative weight updates, and various notions of matrix rank. Finally, the workshop will cover related areas where these kinds of techniques are employed: sparsification, discrepancy and hyperbolic/real stable polynomials.
For more information, please visit: https://simons.berkeley.edu/workshops...
These presentations were supported in part by an award from the Simons Foundation.
1 1:07:05 Constructing Extended Formulations
2 33:35 Small (Explicit) Extended Formulation for Knapsack Cover Inequalities from Monotone Circuits
3 27:36 Computing the Nucleolus of Weighted Cooperative Matching Games in Polynomial Time
4 29:09 A Lower Bound on the Positive Semidefinite Rank of Convex Bodies
5 31:57 Bounds for Matrix Factorization Ranks via Semidefinite Programming and...
6 29:38 Using Symmetry in Semidefinite Programming
7 30:26 Sparse Polynomial Interpolation: Compressed Sensing, Super-resolution, or Prony?
8 1:02:23 Pseudocalibration and SoS Lower Bounds
9 27:20 Weak Decoupling, Polynomial Folds, and Approximate Optimization over the Sphere
10 30:10 Spectral Aspects of Symmetric Matrix Signings
11 31:52 From Weak to Strong LP Gaps for all CSPs
12 31:55 LP, SOCP, and Optimization-Free Approaches to Polynomial Optimization
13 28:54 Matrix Completion and Sums of Squares
14 33:21 SOS and the Dreaded Bit-complexity
15 29:13 Maximizing Sub-determinants and Connections to Permanents and Inequalities on Stable Polynomials
16 32:20 Spectrahedra and Directional Derivatives of Determinants
17 1:06:43 Counting and Optimization Using Stable Polynomials
18 58:02 Independent Set Polytopes: Query vs. Communication Perspective
19 39:03 Approximating Rectangles by Juntas and Weakly-Exponential Lower Bounds for LP Relaxations of CSPs
20 1:04:06 From Proofs to Algorithms for Machine Learning Problems
21 26:59 Nonnegative Polynomials, Nonconvex Polynomial Optimization, and Applications to Learning
22 39:00 Fast Spectral Algorithms from Sum-of-Squares Analyses
23 30:15 Computing the Independence Polynomial in Shearer's Region for the LLL
24 35:40 Quantum Entanglement, Sum of Squares, and the Log Rank Conjecture
Real-Time Decision Making Boot Camp
source: Simons Institute 2018年1月22日
The Boot Camp is intended to acquaint program participants with the key themes of the program, including applications of control theory, machine learning, streaming algorithms and sub-linear algorithms to robotic astronomy, early earthquake warning, the Large Hadron Collider, urban transportation, and control of the energy grid. It will consist of five days of tutorial presentations from the following speakers:
Peter Nugent (Lawrence Berkeley National Laboratory): Supernovae and the Era of Synoptic Surveys
Danny Goldstein (UC Berkeley): Supernova Detection
Richard Allen (UC Berkeley) and Qingkai Kong (UC Berkeley): Earthquake Early Warning
Richard Murray (Caltech): Feedback Control Theory: Architectures and Tools for Real-Time Decision Making
Steven Low (Caltech): The Flow of Power
Sean Meyn (University of Florida): TBA
Kameshwar Poolla (UC Berkeley): The Flow of Money
Sid Banerjee (Cornell University): Ridesharing
Balaji Prabhakar (Stanford University): TBA
David Shmoys (Cornell University): Operating Bike-sharing Systems - A Case Study in Real-time Decision-making
Anna Gilbert (University of Michigan): Introduction to Streaming Algorithms
Ronitt Rubinfeld (MIT): Sub-Linear Time Algorithms: Fast, Cheap and (Only a Little) Out of Control
Evdokia Nikolova (UT Austin): A Brief Introduction to Algorithms, Game Theory and Risk-averse Decision Making
Harvey Newman (Caltech): Physics at the Large Hadron Collider: A New Window on Matter, Spacetime and the Univers
For more information, please visit: https://simons.berkeley.edu/workshops...
These presentations were supported in part by an award from the Simons Foundation.
1 1:13:38 Supernova Detection
2 1:12:49 Supernovae and the Era of Synoptic Surveys
3 1:13:01 Earthquake Early Warning: Streaming Data to Reduce the Loss
4 1:09:46 Earthquake Early Warning: MyShake - Building a Global Seismic Network Using Your Smartphones
5 1:07:57 Understanding and Shaping Urban Mobility Using Dollars and Data I
6 1:02:37 Understanding and Shaping Urban Mobility Using Dollars and Data II
7 1:31:38 Operating Bike-Sharing Systems – A Case Study in Real-Time Decision Making
8 1:38:25 Ridesharing
9 1:11:50 A Brief Introduction to Algorithms, Game Theory and Risk-Averse Decision Making
10 1:36:31 Sub-Linear Time Algorithms: Fast, Cheap and (Only a Little) Out of Control
11 1:00:49 Feedback Control Theory: Architectures and Tools for Real-Time Decision Making I
12 1:17:14 Feedback Control Theory: Architectures and Tools for Real-Time Decision Making II
13 1:02:27 The Flow of Power (Part I: Basic Concepts and Models)
14 1:04:30 The Flow of Power (Part II: Power Flow Solutions and Optimization)
15 1:30:23 The Flow of Information
16 1:27:33 The Flow of Money
17 1:30:39 Physics at the Large Hadron Collider: A New Window on Matter, Spacetime and the Universe I
18 59:31 Physics at the Large Hadron Collider: A New Window on Matter, Spacetime and the Universe II
The Brain and Computation Boot Camp
source: Simons Institute 2018年1月16日
The Boot Camp is intended to acquaint program participants with the key themes of the program. It will consist of four days of tutorial presentations, each with ample time for questions and discussion, as follows:
Anton Arkhipov (Allen Institute for Brain Science) - Allen Institute Large-scale Datasets and Modeling Tools
Jeff Lichtman (Harvard University) - Mapping Neural Connections and Their Development
Wolfgang Maass (Technische Universität Graz) - Computation in Networks of Neurons in the Brain
Bartlett Mel (USC) - Biophysics of Computation
Bruno Olshausen (UC Berkeley) - Neural Mechanisms of Vision
Christos Papadimitriou (UC Berkeley) - Theory of Computation
Peggy Seriès (University of Edinburgh) - Bayesian Theories of Perception and Cognition
Nir Shavit (MIT) - High Throughput Connectomics
Murray Sherman (University of Chicago) - Thalamocortical System
Lena Ting (Emory University and Georgia Institute of Technology) - Neuromechanics and Sensorimotor Control
Santosh Vempala (Georgia Institute of Technology) - Unsupervised Learning
For more information please visit: https://simons.berkeley.edu/workshops...
These presentations were supported in part by an award from the Simons Foundation.
1 1:02:22 Mapping Neural Connections and Their Development I
2 1:12:16 Mapping Neural Connections and Their Development II
3 1:31:51 High Throughput Connectomics
4 1:36:16 Allen Institute Large-scale Datasets and Modeling Tools
5 1:02:26 Biophysics of Computation I
6 1:07:51 Biophysics of Computation II
7 1:00:44 Computation in Networks of Neurons in the Brain I
8 1:13:11 Computation in Networks of Neurons in the Brain II
9 1:02:02 Thalamocortical System I
10 1:04:07 Thalamocortical System II
11 1:31:22 The Brain and Body Compute Together: Neuromechanics and Sensorimotor Control
12 1:33:35 Neural Mechanisms of Vision
13 1:00:32 Theory of Computation I
14 1:07:11 Theory of Computation II
15 1:31:08 Unsupervised Learning
16 1:37:39 Bayesian Theories of Perception and Cognition
CEMRACS 2017 - Seminars
source: Centre International de Rencontres Mathématiques 2017年9月8日
1 56:54 Jorge P. Zubelli: Project Evaluation under Uncertainty
2 55:28 Stefano De Marco: Some asymptotic results about American options and volativity
3 53:15 Denis Belomestny: Projected particle methods for solving McKean Vlasov SDEs
4 56:03 Francisco José Silva Álvarez: On the discretization of some nonlinear Fokker-Planck-Kolmogorov ...
5 49:25 Emmanuel Gobet : Forward and backward simulation of Euler scheme
6 43:14 Ludovic Goudenège: Splitting algorithm for nested events
7 47:29 Mathieu Laurière: Mean field type control with congestion
8 47:01 Laurent Mertz: Stochastic variational inequalities for random mechanics
Jean-Morlet Chair - Khanin/Shlosman - 1st Semester 2017
source: Centre International de Rencontres Mathématiques 2017年8月1日
1 18:53 Interview at Cirm: Konstantin Khanin
2 54:31 Martin Hairer: Weak universality of the KPZ equation with arbitrary nonlinearities
3 19:36 Interview at Cirm: Martin Hairer
4 4:29 Martin Hairer ITV Behind the scenes
5 1:31:48 Timo Seppäläinen: Variational formulas, Busemann functions, and fluctuation exponents - Part 1
6 1:38:37 Timo Seppäläinen: Variational formulas, Busemann functions, and fluctuation exponents - Part 2
7 1:34:59 Timo Seppäläinen: Variational formulas, Busemann functions, and fluctuation exponents - Part 3
8 50:12 Vadim Gorin: Tilings and non-intersecting paths beyond integrable cases
9 46:15 Dmitry Ioffe: Low temperature interfaces and level lines in the critical prewetting regime
10 52:43 Fabio Toninelli: A second growth model in the Anisotropic KPZ class
11 44:50 Jinho Baik: Multi-time distribution of periodic TASEP
12 1:29:52 Daniel Remenik: The KPZ fixed point - Part 2
13 1:37:55 Daniel Remenik: The KPZ fixed point - Part 1
Subscribe to:
Posts (Atom)