Loriano Mancini (EPF Lausanne)
10:00 • EPF Lausanne, Salle BI A0 448
Prof. Dr. Ana Caraiani (University of Bonn)
Galois representations and torsion classes
13:15 • UZH Irchel, Winterthurerstrasse 190, Zürich, Building Y27, Room H 25
Georgios Pitselis (University of Piraeus, Greece)
Another Look at Risk Measures in a Credibility Framework abstract
Abstract:
Here, we introduce a new type of risk measures, the credible risk measures, in order to capture the risk of an individual contract (or financial portfolio) as well as the industry risk consisting of several, similar but not identical, contracts (or financial portfolios). We show how risk measures can be embedded within the framework of credibility theory in the regression case. Examples are given based on the Fama/French data.
14:00 • EPF Lausanne, Quartier UNIL - Dorigny, Salle Extranef 109
Loriano Mancini (EPF Lausanne)
14:00 • EPF Lausanne, Salle BI A0 448
Shahar Mendelson (Australian National University, Canberra, and Technion, Haifa)
The small-ball method and the structure of random coordinate projections abstract
Abstract:
We study the geometry of the natural function class extension of a random projection of a subset of $R^d$: for a class of functions $F$ defined on the probability space $(\\Omega,\\mu)$ and an iid sample X_1,...,X_N with each of the $X_i$\'s distributed according to $\\mu$, the corresponding coordinate projection of $F$ is the set $\\{ (f(X_1),....,f(X_N)) : f \\in F\\} \\subset R^N$. We explain how structural information on such random sets can be derived and then used to address various questions in high dimensional statistics (e.g. regression problems), high dimensional probability (e.g., the extremal singular values of certain random matrices) and high dimensional geometry (e.g., Dvoretzky type theorems). Our focus is on results that are (almost) universally true, with minimal assumptions on the class $F$; these results are established using the recently introduced small-ball method.
15:15 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room G 19.2
Kevin Schnelli (KTH, Stockholm)
Addition of random matrices abstract
Abstract:
Describing the eigenvalue distribution of the sum of two general Hermitian matrices is basic question going back to Weyl. If the matrices have high dimensionality and are in general position in the sense that one of them is conjugated by a random Haar unitary matrix, the eigen-value distribution of the sum is given by the free additive convolution of the respective spectral distributions. This result was obtained by Voiculescu on the macroscopic scale. In the rst part of this talk, I will show that this convergence holds on the microscopic scale all the way down to the eigenvalue spacing. In the second part of this talk, I will discuss how these results can be used to derive the optimal convergence rate in the Single Ring Theorem. Joint work with Zhigang Bao and Laszlo Erdos.
15:15 • Université de Genève, Section Mathématiques, 2-4, rue du Lièvre, Salle 623
Dr. Doris Hein (Universität Freiburg)
Local invariant Morse theory and applications in Hamiltonian dynamics abstract
Abstract:
Local homology is a useful tool to study periodic orbits. For example, the key to the existence of infinitely many periodic orbits of Hamiltonian systems are properties of the local Floer homology of one special orbit.I will discuss a discrete version of this invariant constructed using local invariant Morse homology of a discrete action function. The construction is very geometric and relies on a hands-on description of invariant local Morse homology.The resulting local homology can be interpreted as an invariant of germs of Hamiltonian systems or of closed Reeb orbits. It has properties similar to those of local Floer homology in the symplectic setting and probably similar applications in dynamics.
15:15 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room G 43
Mladen Kolar (The University of Chicago)
Some Recent Advances in Scalable Optimization abstract
Abstract:
In this talk, I will present two recent ideas that can help solvelarge scale optimization problems. In the first part, I will present amethod for solving an ell-1 penalized linear and logistic regressionproblems where data are distributed across many machines. In such ascenario it is computationally expensive to communicate informationbetween machines. Our proposed method requires a small number ofrounds of communication to achieve the optimal error bound. Withineach round, every machine only communicates a local gradient to thecentral machine and the central machine solves a ell-1 penalizedshifter linear or logistic regression. In the second part, I willdiscuss usage of sketching as a way to solve linear and logisticregression problems with large sample size and many dimensions. Thiswork is aimed at solving large scale optimization procedures on asingle machine, while the extension to a distributed setting is workin progress.
16:30 • ETH Zentrum, Rämistrasse 101, Zürich, Building HG, Room G 19.2