See all upcoming seminars in LäsIT and seminar web pages at the homepage for the PhD studentseminars, TDB, Vi2, Theory and Applications Seminars (TAS) @ UpMARC., Department of Mathematics and The Stockholm Logic Seminar.
|InfoLab fika meeting|
|Mikaela Micho: Bayesian Stochastic Blockmodeling|
Location: ITC 19204, Time: 14:15-15:00
We meet in room 204 (upper floor), House 19 where our MSc student Mikaela Micho will present bayesian stochastic blockmodeling (SBM), a family of methods to find communities in complex networks. After the presentation we will discuss how SBM can be extended to find communities combining the network structure with other attributes of the nodes. Then fika.
|Igor Tominec: A radial basis function finite difference method with improved stability properties|
Location: ITC 2345, Time: 13:15
AbstractMethods for solving elliptic partial differential equations based on radial basis functions (RBFs) in a collocation setup are prone to instabilities in the presence of Neumann boundary conditions. This is unfortunate since such conditions are important to use when for example constructing a mechanical model of the human diaphragm. We solve this issue by formulating the RBF finite difference method in a least-squares setup instead. The change of the residual minimization framework allowed us to prove convergence under node refinement and to further develop the method to support a trial space which does not conform to the geometry on which the differential equation is solved.
|Disputation | Dissertation|
|Ida Löscher: Aiming at Moving Targets: Applying Cognitive Work Analysis to Work Domains in Transition|
Location: ITC 2446, Time: 13:15
Ida Löscher will present and defend her PhD thesis: Aiming at Moving Targets: Applying Cognitive Work Analysis to Work Domains in Transition.
|Licentiatseminarium | Licentia|
|Christos Sakalis: Securing the Memory Hierarchy from Speculative Side-Channel Attacks|
Location: ITC 1211, Time: 13:00
External reviewer: Sven Karlsson
Modern high-performance CPUs depend on speculative out-of-order execution in order to offer high performance while also remaining energy efficient. However, with the introduction of Meltdown and Spectre in the beginning of 2018, speculative execution has been under attack. These attacks, and the many that followed, take advantage of the unchecked nature of speculative execution and the microarchitectural changes it causes in order to mount speculative side-channel attacks. Such attacks can bypass software and hardware barriers and gain access to sensitive information while remaining invisible to the application.
|Ashkan Panahi: Sum of Norms: A Convex Optimization Approach to Clustering|
Location: Ångström 72121, Time: 14:15-15:00
AbstractVector clustering is one of the most popular problems of unsupervised machine learning with the K- means algorithm as its highly celebrated solution. However, K-means reportedly falls short of reliability on account of its unpredictable convergence properties. In this talk, we present an alternative convex optimization framework for clustering, known as the sum of norms (SON) approach with provable convergence properties. We show that SON possesses multiple desirable properties, such as guaranteed performance in par with the known results for K-means, and convenient implementation through incremental optimization techniques with extremely low-cost iterations. We further discuss the application of SON as a regularizer in other ML tasks such as optimal transport.
Bio:Ashkan Panahi is an assistant professor at the Computer Science and Engineering Department at Chalmers University, Sweden. He respectively received his BSc and MSc degrees in electrical and communication systems engineering from Iran University of Science and Technology (IUST) (2008) and Chalmers University (2010). He also received his PhD degree in Signal Processing from Electrical Engineering Department at Chalmers (2015). He has held multiple other research positions such as visiting research student at California Institute of Technology (Caltech 2014), U.S. National Research Council Research Associate (2016-2018) and Postdoctoral Researcher at North Carolina State University (2018-2019). His research interest spans a broad range of topics in the theory of machine learning and data processing, including optimization algorithms, statistical analysis and probabilistic methods, compressed sensing, and statistical detection and estimation theory.
|Disputation | Dissertation|
|Jing Liu: Towards Fast and Robust Algorithms in Flash X-ray single-particle Imaging|
Location: BMC B41, Time: 9:15
Jing Liu will present and defend her PhD thesis: Towards Fast and Robust Algorithms in Flash X-ray single-particle Imaging.
See also the list of all upcoming seminars.
Internal seminars. Lecturers may be either internal or external.