CMU Mathematics Department Colloquia

FALL 2018

Organizer

Meeting Times

Refreshments take place 30 minutes prior to the talk.
Typical Colloquium Talks are Thursday, 4:00–4:50pm, in Pearce 227. Refreshments are in Pearce 216.
There might be some talks scheduled on a different day and time. The following table gives the most accurate information for each event.

Schedule

Date Speaker Title (Scroll down for Abstract) Remark
8/30/2018
9/6/2018 Department Meeting Department Meeting
9/13/2018 Botong Wang (University of Wisconsin -- Madison) The Algebraic Geometry in the "Top-Heavy" Conjecture
9/18/2018
Tuesday
Dr. Dan Wardrop (Central Michigan Alumni) A Senior Statistician's Career Path in Statistics and Analytics
9/20/2018 Kyoung-Jae Won (University of Copenhagen, Denmark)
Biotech Research and Innovation Centre (BRIC)
A New Clustering Approach Using Modality in Single Cell RNAseq Data Outperforms Classical Hierarchical Clustering
9/27/2018 Department Meeting Department Meeting
10/1/2018
Monday
4 pm
Anna Sfard (University of Haifa)
Fleming Lecture Series
Mathematics Learning: Does Language Make a Difference? BIO 1010
10/2/2018
Tuesday
Anna Sfard (University of Haifa)
Fleming Lecture Series
What Is Mathematics and Why Does It Matter? PE 127
10/4/2018
10/11/2018
10/18/2018 Philip Hackney (University of Louisiana) On Operads and Homotopy
10/23/2018
Tuesday
Yeon Ju Lee (Korean University Sejong Campus) Data Adaptive Least Squares Methods For Image Processing
10/25/2018 Department Meeting Department Meeting
11/1/2018
11/8/2018 Ji Zhu (University of Michigan) Estimating Network Edge Probabilities By Neighborhood Smoothing
11/15/2018
11/29/2018 Department Meeting Department Meeting
12/6/2018

Abstracts

Speaker: Botong Wang (September 13)
Title: The Algebraic Geometry in the "Top-Heavy" Conjecture
Abstract: A theorem of de Bruijn and Erdos states that n points in the plane determines at least n lines, unless all points are in one line. A higher dimensional generalization of this theorem was conjectured by Dowling and Wilson in 1974. We will discuss a proof the the conjecture by exploring the algebraic geometry behind the combinatorial constants. This is a joint work with June Huh.

Speaker: Dan Wardrop (September 18)
Title: A Senior statistician's Career Path in Statistics and Analytics
Abstract: This non-technical, and interactive, talk is aimed at undergraduate and graduate students. The intent, unashamedly, is to convince one or more of you to consider a graduate education is applied statistics. I will review my own career path and describe how my undergraduate and graduate experiences helped lead to a varied and rewarding career at Royal Dutch Shell. Come prepared with questions that you have regarding a career in statistics.
Bio: This non-technical, and interactive, talk is aimed at undergraduate and graduate students. The intent, unashamedly, is to convince one or more of you to consider a graduate education is applied statistics. I will review my own career path and describe how my undergraduate and graduate experiences helped lead to a varied and rewarding career at Royal Dutch Shell. Come prepared with questions that you have regarding a career in statistics.

Speaker: Kyoung-Jae Won (September 20)
Title: A New Clustering Approach Using Modality in Single Cell RNAseq Data Outperforms Classical Hierarchical Clustering
Abstract: Single-cell RNA sequencing (scRNA-seq) is a powerful tool to study heterogeneity and dynamic changes in cell populations. Clustering scRNA-seq is essential in identifying new cell types and studying their characteristics. We develop CellBIC (single Cell BImodal Clustering) to cluster scRNA-seq data based on modality in the gene expression distribution. Compared with classical bottom-up approaches that rely on a distance metric, CellBIC performs hierarchical clustering in a top-down manner. CellBIC outperformed the bottom-up hierarchical clustering approach and other recently developed clustering algorithms while maintaining the hierarchical structure of cells.
I will also discuss how to increase scalability to handle large-scale scRNAseq data.

Speaker: Anna Sfard (October 1 & 2)
Title: (October 1) Mathematics Learning: Does Language Make a Difference?
Abstract: This talk is accessible to a wide audience with curious minds. The focus of this talk is on the impact of language on mathematics learning.
Title: (October 2) What Is Mathematics And Why Does It Matter?
Abstract: This talk will be on the questions of what it means to think mathematically and of its significance for how mathematics is being taught and learned.

Speaker: Philip Hackney (October 18)
Title: On Operads and Homotopy
Abstract: Operads are algebraic gadgets which control various types of algebras. For example, there is an operad L so that the set of actions of L on a vector space V is in bijection with the set of Lie algebra structures on V (similarly, there are operads controlling associative and commutative algebra structures). Variations on the concept of operad allow one to model other types of structures, such as Hopf algebras and maps of associative algebras. We will give examples that are particularly relevant to homotopy theory and also discuss homotopy-coherent versions of operads.

Speaker: Yeon Ju Lee (October 23)
Title: Data Adaptive Least Squares Methods For Image Processing
Abstract: The importance and utilization of data approximation for signal and image processing are rapidly increasing in IT related industries and various research fields. In order to process data having different shapes and characteristics in various fields, it is necessary to accurately analyze characteristics of a given data and develop a data-adaptive approximation technique. We introduce a data-adaptive least squares method that approximates the data by reflecting the characteristics of the data. We develop mathematical theory and solve the problem of actual image data processing, especially noise reduction.

Speaker: Ji Zhu (November 8)
Title: Estimating Network Edge Probabilities by Neighborhood Smoothing
Abstract: The problem of estimating probabilities of network edges from the observed adjacency matrix has important applications to predicting missing links and network denoising. It has usually been addressed by estimating the graphon, a function that determines the matrix of edge probabilities, but is ill-defined without strong assumptions on the network structure. Here we propose a novel computationally efficient method based on neighborhood smoothing to estimate the expectation of the adjacency matrix directly, without making the strong structural assumptions graphon estimation requires. The neighborhood smoothing method requires little tuning, has a competitive mean-squared error rate, and outperforms many benchmark methods on the task of link prediction in both simulated and real networks. This is joint work with Yuan Zhang and Elizaveta Levina.
Bio: Professor Ji Zhu obtained his B.Sc. in Physics from Peking University in 1996 and his Ph.D. in Statistics from Stanford University in 2003 with the dissertation advisor Professor Trevor Hastie. His research interests include statistical learning, high-dimensional data and statistical network analysis. He is also interested in applications in medicine, computational biology, engineering, physics and business. Professor Zhu received a CAREER award from the NSF in 2008 and was elected a member of ISI in 2010, a fellow of ASA in 2013, and a fellow of IMS in 2015.

Past Department Colloquia: Spring 2018 Fall 2017 Spring 2017 Fall 2016

Look for Colloquia Archives for past activities