Day: October 10, 2019

MTRI to Present Research Seminars October 14

The Institute of Computing and Cybersystems will present four brief seminars by researchers from the Michigan Tech Research Institute (MTRI) on Monday, October 14, 2019, 11:00 a.m. to 12:00 p.m., in EERC 122.  MTRI research and outreach focuses on the development of technology to sense and understand natural and manmade environments.

Sarah Kitchen is a mathematician with background in algebraic geometry and representation theory. Her recent research interests include algebraic structures underlying optimization problems and applications of emerging statistical tools to signal processing and source separation problems. Her talk, “Collaborative Autonomy,” will discuss some considerations in centralized, semi-centralized, and decentralized decision-making methods for autonomous systems.

Susan Janiszewski is a mathematician specializing in graph theory and combinatorics. Her research interests lie in applying concepts from discrete mathematics to machine learning, computer vision, and natural language processing. Her talk, “Combining Natural Language Processing and Scalable Graph Analytics,” takes up the fast-growing field of Natural Language Processing (NLP), i.e. the development of algorithms to process large amounts of textual data. Janiszewski will discuss ways to combine common NLP and graph theoretic algorithms in a scalable manner for the purpose of creating overarching computational systems such as recommendation engines or machine common sense capabilities.

Joel LeBlanc has 10 years of experience in statistical signal processing. His research interests include information theoretic approaches to inverse imaging, and computational techniques for solving large inverse problems. LeBlanc’s talk, “Testing for Local Minima of the Likelihood Using Reparameterized Embeddings,” addresses the question: “Given a local maximum of a non-linear and non-convex log-likelihood equation, how should one test for global convergence?” LeBlanc will discuss a new strategy for identifying globally optimal solutions using standard gradient-based optimization techniques.

Meryl Spencer is a physicist with a background in complex systems and network theory. Her research interests include machine learning for image processing, applications of graph algorithms, and self-organization. Her talk, “Computational modeling of collaborative multiagent systems,” will discuss her previous work on modeling self organization in cellular networks, and some areas of interest for future work.

Download the event flyer.

Anna Little to Present Talk October 18, 1 p.m.

Anna Little

Dr. Anna Little, a postdoc in the Department of Computational Mathematics, Science, and Engineering at Michigan State University, will present her talk, “Robust Statistical Procedures for Clustering in High Dimensions,” on Friday, October 18, 2019, at 1:00 p.m., in Fisher Hall Room 327B.

Dr. Little completed a PhD in mathematics at Duke University in 2011. She has been at Michigan State since 2018.  Visit her website at

Lecture Abstract: This talk addresses multiple topics related to robust statistical procedures for clustering in high dimensions, including path-based spectral clustering (a new method), classical multidimensional scaling (an old method), and clustering in signal processing. Path-based spectral clustering is a novel approach which combines a data driven metric with graph-based clustering. Using a data driven metric allows for fast algorithms and strong theoretical guarantees when clusters concentrate around low-dimensional sets.

Another approach to high-dimensional clustering is classical multidimensional scaling (CMDS), a dimension reduction technique widely popular across disciplines due to its simplicity and generality. CMDS followed by a simple clustering algorithm can exactly recover all cluster labels with high probability when the signal to noise ratio is high enough. However, scaling conditions become increasingly restrictive as the ambient dimension increases, illustrating the need for robust unbiasing procedures in high dimensions.  Clustering in signal
processing is the final topic; in this context each data point corresponds to a corrupted signal. The classic multireference alignment problem is generalized to include random dilation in addition to random translation and additive noise, and a wavelet based approach is used to define an unbiased representation of the target signal(s) which is robust to high frequency perturbations.

Download the event flyer.