Department of Mathematics

Search

en

DACO Seminar

×

Modal title

Modal content

Pleasesubscribe hereif you would you like to be notified about these presentations via e-mail. Moreover you cansubscribeto the iCal/ics Calender.

Autumn Semester 2023

Date / Time Speaker Title Location
*3 October 2023
15:05-16:00		 Gabriel Arpino
University of Cambridge
Event Details

DACO Seminar

Title Statistical-Computational Tradeoffs in Mixed Sparse Linear Regression
Speaker, Affiliation Gabriel Arpino,University of Cambridge
Date, Time 3 October 2023, 15:05-16:00
Location HG G 19.2
Abstract We consider the problem of mixed sparse linear regression with two components, where two sparse signals are observed through n unlabelled noisy linear measurements. Prior work has shown that the problem suffers from a significant statistical-to-computational gap, resembling other computationally challenging high-dimensional inference problems such as Sparse PCA and Robust Sparse Mean Estimation. We establish the existence of a more extensive computational barrier for this problem through the method of low-degree polynomials, but show that the problem is computationally hard only in a very narrow symmetric parameter regime. We identify smooth information-computation tradeoffs in this problem and prove that a simple linear-time algorithm succeeds outside of the narrow hard regime. To the best of our knowledge, this is the first thorough study of the interplay between mixture symmetry, signal sparsity, and their joint impact on the computational hardness of mixed sparse linear regression. This is joint work with Ramji Venkataramanan. https://proceedings.mlr.press/v195/arpino23a.html.
Statistical-Computational Tradeoffs in Mixed Sparse Linear Regressionread_more
HG G 19.2
17 October 2023
14:15-15:15		 Dr. Andrew McRae
EPFL
Event Details

DACO Seminar

Title Title T.B.A.
Speaker, Affiliation Dr. Andrew McRae,EPFL
Date, Time 17 October 2023, 14:15-15:15
Location HG G 19.1
Title T.B.A.
HG G 19.1
24 October 2023
14:15-15:15		 Dr. Daria Tieplova
ICTP, Trieste
Event Details

DACO Seminar

Title Fundamental limits of overparametrized shallow neural networks for supervised learning
Speaker, Affiliation Dr. Daria Tieplova,ICTP, Trieste
Date, Time 24 October 2023, 14:15-15:15
Location HG G 19.1
Abstract I will discuss the joint work done with Francesco Camilli and Jean Barbier concerning an information-theoretical analysis of a two-layer neural network trained from input-output pairs generated by a teacher network with matching architecture, in overparametrized regimes. Our results come in the form of bounds relating i) the mutual information between training data and network weights, or ii) the Bayes-optimal generalization error, to the same quantities but for a simpler (generalized) linear model for which explicit expressions are rigorously known. Our bounds, which are expressed in terms of the number of training samples, input dimension and number of hidden units, thus yield fundamental performance limits for any neural network (and actually any learning procedure) trained from limited data generated according to our two-layer teacher neural network model. The proof relies on rigorous tools from spin glasses and is guided by ``Gaussian equivalence principles'' lying at the core of numerous recent analyses of neural networks. With respect to the existing literature, which is either non-rigorous or restricted to the case of the learning of the readout weights only, our results are information-theoretic (i.e. are not specific to any learning algorithm) and, importantly, cover a setting where all the network parameters are trained.
Fundamental limits of overparametrized shallow neural networks for supervised learningread_more
HG G 19.1
31 October 2023
14:15-15:15		 Prof. Dr. Ivan Dokmanić
University of Basel
Event Details

DACO Seminar

Title Statistical Mechanics of Generalization In Graph Convolution Networks
Speaker, Affiliation Prof. Dr. Ivan Dokmanić,University of Basel
Date, Time 31 October 2023, 14:15-15:15
Location HG G 19.1
Statistical Mechanics of Generalization In Graph Convolution Networks
HG G 19.1

Notes: thehighlighted eventmarks the next occurring event and events marked with an asterisk (*) indicate that the time and/or location are different from the usual time and/or location.

Archive:HS 23FS 23HS 22FS 22HS 21FS 21HS 20FS 20

JavaScript has been disabled in your browser