Probabilistic
techniques &
algorithms
Overview
The design of fast algorithms plays a crucial role in various branches of applied mathematics and engineering. Often an algorithm's efficiency can be greatly improved using randomization, at the cost of allowing for a small probability of failure. Examples include the use of random measurements in compressed sensing and randomized algorithms for fast linear algebra. Probabilistic techniques can also be of use when the complexity of the data prevents the efficient use of adaptive algorithms. An example is given by nearest neighbor searches in massive high-dimensional data sets. Applying a random projection can make the calculations more efficient while, with high probability, introducing only a small error.
The workshop aims to bring together experts from different fields in applied mathematics, computer science and engineering whose research involves probabilistic techniques and randomized algorithms. The goal is to put recent results into perspective, encourage collaboration, and to discuss how the probabilistic viewpoint can find new fields of application.
Limited funding for local support will be provided for young participants. Funding decisions have been made.
This conference is sponsored by The Harrington Fellows Program.
The workshop will take place at the University Teaching Center (UTC) in Room 1.130 on the UT campus.
Program
(Full schedule here. )
Friday, April 6, 2012
8.50 - 9.00 | Opening | |
9.00 - 9.25 | Adam Klivans | Learning convex sets |
9.30 - 9.55 | Mark Iwen | A symbol-based barcode decoding algorithm |
10.00 - 10.25 | Lorenzo Rosasco | Unsupervised learning and regularization |
Coffee break | ||
11.00 - 11.25 | Pablo Parrilo | TBA |
11.30 - 11.55 | Michael Mahoney | Approximate computation and implicit regularization in large-scale data analysis |
12.00 - 12.25 | Piotr Indyk | Faster algorithms for Sparse Fourier Transform |
Lunch | ||
2.30 - 2.55 | Joel Tropp | How to find a needle in a haystack |
3.00 - 3.25 | Rayan Saab | Sigma delta quantization of Gaussian frame expansions: root-exponential accuracy |
3.30 - 3.55 | Raj Rao Nadakuditi | Some surprises in the estimation of low-rank matrix-valued random variables |
Coffee break | ||
4.30 - 4.55 | Venkat Chandrasekaran | The convex geometry of linear inverse problems |
5.00 - 5.25 | Maryam Fazel | TBA |
5.30 - 5.55 | Ben Recht | System identification with atomic norm regularization |
Workshop dinner |
Saturday, April 7, 2012
9.00 - 9.25 | Yaniv Plan | Structured signal recovery from single-bit measurements |
9.30 - 9.55 | Simon Foucart | Stability and robustness of l1-minimizations with Weibull matrices and redundant dictionaries |
10.00 - 10.25 | Constantine Caramanis | Hypothesis testing on social networks |
Coffee break | ||
11.00 - 11.25 | Nir Ailon | Efficient adaptive querying strategies for clustering and ordering problems |
11.30 - 11.55 | Luis Rademacher | Recent developments in column subset selection and volume sampling |
12.00 - 12.25 | Nikhil Srivastava | Covariance estimation for distributions with 2+epsilon moments |
Lunch | ||
2.30 - 2.55 | Mauro Maggioni | Multiscale geometric methods for noisy point clouds in high dimensions |
3.00 - 3.25 | Laurent Demanet | TBA |
3.30 - 3.55 | Deanna Needell | Robust image recovery via total variation minimization |
Coffee break | ||
4.30 - 4.55 | Sujay Sanghavi | Learning Stochastic Dynamical Systems with Latent Variables |
5.00 - 5.25 | Massimo Fornasier | Learning functions of few arbitrary linear parameters in high dimensions |
5.30 - 5.40 | Closing |
Accommodation
A block of rooms have been reserved at the Extended Stay America in downtown Austin. Directions from the Extended Stay to the workshop can be found here.Organizing Committee
Rachel WardFelix Krahmer
Holger Rauhut