Events by month

Title: How can we optimize nonsmooth objectives globally?

Speaker: Andreas Griewank Affiliation: Humboldt University, Germany Room: MC 5501

Abstract:

In machine learning objective functions that are only piecewise smooth and should be globally minimized abound. The standard method of dealing with them is to apply a stochastic gradient method disregarding the rare points of nonsmoothness and hoping for the best as far as global optimality of the computed solution is concerned.

Title: Inverses of Trees

Speaker: Krystal Guo Affiliation: Centre de Recherches Mathématiques & Université de Montréal Room: MC 5417

Abstract:

A tree is invertible if and only if it has a perfect matching.

Godsil considers an invertible tree T and finds that the inverse of the adjacency matrix has entries in {0, ±1} and is the signed adjacency matrix of a graph which contains T.

Title: Biometric security: Design, implementation, and challenges

Speaker: Koray Karabina Affiliation: Florida Atlantic University Room: MC 5501

Abstract:

Biometric based cybersecurity technologies offer significant advantages in authentication, identification, and access control mechanisms.

Title: Graph covers with two new eigenvalues

Speaker: Olha Silina Affiliation: University of Waterloo Room: MC 5479

Abstract:

I am going to talk about graph covers. If $Y$ is a cyclic cover of $X$, it turns out that $Y$ has the same spectrum as $X$ plus (possibly) some new eigenvalues. 

Title: OutFn, M_gn and renormalized topological field theory

Speaker: Michael Borinsky Affiliation: Nikhef Room: MC 5417

Abstract:

I will report on recent joint work with Karen Vogtmann on the Euler characteristic of Out(F_n) and the moduli space of graphs.