Top

Colloquia

Professor Alexandre Xavier Falcão 9/10/2014

Image Segmentation using
The Image Foresting Transform

Date: 9/10/2014
Time: 3:30PM-4:30PM
Place: 315 Armstrong Hall

Professor Alexandre Xavier Falcão

Abstract:
Image segmentation is a challenging task that consists of an image
partition into either superpixels (regions) that include the
delineation of the desired object borders, or define the precise
spatial extent of such objects in the image. The image foresting
transform (IFT) provides a unified framework to the design of
operators based on optimum connectivity between image elements
(pixels, superpixels, or their components). Its applications include
filtering, segmentation, distance transforms, skeletonization, shape
description, clustering, and classification.

This lecture presents a short overview on the IFT in order to discuss
several aspects related to boundary-based, region-based, interactive,
and automated image segmentation. The IFT algorithms rely on the
suitable choice of an adjacency relation and a connectivity
function. The adjacency relation defines an image graph, whose nodes
are the image elements and arcs connect the adjacent ones. The
connectivity function assigns a value to any path in the graph,
including trivial ones formed by a single node. The maxima (minima) of
the trivial connectivity map are called "seeds" --- they compete among
themselves to conquer their most strongly connected nodes such that
the image is partitioned into an optimum-path forest rooted at the
winner seeds. The image operators result from the attributes of the
forest (optimum paths, their connectivity values, root labels).

The lecture shows how to extend the IFT to the design of pattern
classifiers and discusses connectivity functions for boundary-based
and region-based segmentation, seed selection and imposition, oriented
boundaries and regions, and the incorporation of object information
(texture and shape models) into the interactive and automated
segmentation processes. Most IFT-based methods execute in time
proportional to the number of nodes (linear or sublinear time). The
lecture discusses how to correct segmentation errors in sublinear time
(without starting the process from the beginning, even when the image
was processed by other segmentation approach) and how to obtain smooth
object boundaries without loosing consistency in segmentation (i.e.,
the nodes remain connected to their respective roots). Finally, it
concludes with the current research on IFT-based image segmentation.

Short Biography

Alexandre Xavier Falcão is professor at the Institute of Computing,
University of Campinas (UNICAMP), Brazil. He received a B.Sc. in
Electrical Engineering from the Federal University of Pernambuco,
Brazil, in 1988. He has worked in biomedical image processing,
visualization, and analysis since 1991. In 1993, he received a
M.Sc. in Electrical Engineering from UNICAMP. During 1994-1996, he
worked with the Medical Image Processing Group at the Department of
Radiology, University of Pennsylvania, USA, on interactive image
segmentation for his doctorate. He got his doctorate in Electrical
Engineering from UNICAMP in 1996. In 1997, he worked in a research
center (CPqD-TELEBRAS) developing methods for video quality
assessment. His experience as professor of Computer Science started in
1998 at UNICAMP. His main research interests include graph algorithms
for image processing, image segmentation, volume visualization,
content-based image retrieval, mathematical morphology, digital video
processing, remote sensing image analysis, machine learning, pattern
recognition, and biomedical image analysis.

Date, Location: 
2014-09-10

Dr. Pete Donnell 9/9/14

Monotonicity, nonexpansivity
and chemical reaction networks

Date: 9/9/2014
Time: 3:30PM-4:30PM
Place: 315 Armstrong Hall

Dr. Pete Donnell

Abstract:
There is currently significant research interest into chemical reaction networks (CRNs), largely due to their importance in biochemistry. CRNs are commonly modelled as ODEs. Given an ODE representing a CRN, characterising its possible asymptotic behaviour is in general a difficult task. A significant proportion of real world CRNs exhibit very simple behaviour, for example global convergence to a unique steady state. However, more exotic CRNs such as the Belousov-Zhabotinsky reactions, which can have a stable nontrivial periodic orbit, are not uncommon.

In this talk I will give a brief overview of two areas of theory, monotonicity and nonexpansivity, that can be used to constrain the possible dynamics of an ODE. It is known that a monotone ODE cannot have stable periodic orbits. If a monotone ODE also has a linear first integral, it is possible in some cases to show that it is nonexpansive and that every bounded trajectory converges to a steady state. Examples drawn from chemical reaction network theory will be presented as applications of this result.

Date, Location: 
2014-09-09

Professor Andreas Brandstädt 9/3/2014

The complexity of the efficient
domination problem

Date: 9/3/2014
Time: 4:00PM-5:00PM
Place: 315 Armstrong Hall

Professor Andreas Brandstädt

Abstract:
Packing and covering problems in graphs and hypergraphs and their relationship
belong to the most fundamental topics in combinatorics and graph algorithms and
have a wide spectrum of applications in computer science, operations research and
many other fields.
Recently, there has been an increasing interest in graph and hypergraph problems
combining packing and covering properties, and the NP-complete Exact Cover problem
on hypergraphs (asking for a collection of hyperedges covering each vertex exactly
once) is a good example for this. Closely related graph problems are the Efficient
Domination (ED) and Efficient Edge Domination (EED) problems; the ED problem
corresponds to the exact cover problem for the closed neighborhoods of a graph,
and the EED problem on a graph G is the ED problem on its line graph L(G).
We give a survey on previous results, describe some connections to other problems
such as Maximum Weight Independent Set and Minimum Weight Dominating
Set, consider some new cases where the problems are efficiently solvable by
using structural properties of graph classes and using closure properties under the
square operation, and we also extend the graph problems to hypergraphs.

Date, Location: 
2014-09-03

Dr. Xin He 7/10/14

Some Graph-Based Models
and Algorithms in Genomics

Date: 7/10/2014
Time: 3:30PM-4:30PM
Place: 315 Armstrong Hall

Dr. Xin He

Abstract:
Graph theory has a number of applications in bioinformatics and genomics
research. In this talk, I will discuss two recent problems that can
formulated in graphical theoretical terms. Next generation sequencing (NGS)
technology greatly reduces the cost of DNA sequencing. Such NGS data
typically consists of a large number of short "reads" of length a few
hundred base pairs or less, and how to assemble these reads to the full
genomic sequences is a challenging combinatorial problem. I will review
some of the algorithms that have been proposed for this problem, which are
essentially finding Hamiltonian or Eulerian path in a graph. In practice,
the problem is made harder by the errors in sequencing, repeats in genomic
sequences, and so on. In the second part of my talk, I will discuss a
problem of mapping genes that contribute to human diseases. Existing
studies tend to analyze one gene at a time through association studies,
which test if the variation of DNA sequences in a gene correlates with the
status of a disease (whether a person is affected or not). Genomic data, on
the other hand, creates a map of how genes are related to each other. I
will discuss some of our attempts at integrating association data (at the
level of individual genes) and gene network data. Such an approach
effectively identifies highly related gene groups (dense subgraphs) that
are most likely to contribute to diseases.

Date, Location: 
2014-07-10

Professor Jerrold Griggs 5/14/2014

Symmetric Venn Diagrams and
Symmetric Chain Decompositions

Date: 5/14/2014
Time: 3:00PM-4:00PM
Place: 315 Armstrong Hall

Jerrold Griggs

Abstract: Here

Date, Location: 
2014-05-14

Professor Gheorghe Craciun 5/2/2014

Date: 5/2/2014
Time: 2:30PM-3:30PM
Place: 315 Armstrong Hall

Gheorghe Craciun

Abstract: Complex interaction networks are present in all areas of biology, and
manifest themselves at very different spatial and temporal scales. Persistence,
permanence and global stability are emergent properties of complex networks, and
play key roles in the dynamics of living systems.
Mathematically, a dynamical system is called persistent if, for all positive
solutions, no variable approaches zero. In addition, for a permanent system, all
variables are uniformly bounded. We describe criteria for persistence and permanence
of solutions, and for global convergence of solutions to an unique equilibrium, in a
manner that is robust with respect to initial conditions and parameter values.
We will also point out some connections to classical problems about general
dynamical systems, such as the construction of invariant sets and Hilbert's 16th
problem.

Date, Location: 
2014-05-02

Professor Xujing Wang 5/1/14

Date: 5/1/2014
Time: 3:30PM-4:30PM
Place: 315 Armstrong Hall

Xujing Wang

Abstract:
Living systems are characterized by complexity in structure and emergent
dynamic orders. In many aspects the onset of a chronic disease resembles a phase
transition in a complex dynamic system: quantitative changes accumulate largely
unnoticed until a critical threshold is reached, which causes abrupt qualitative changes
of the system. This insight can help us identify the key factors driving the
disease, and the critical parameters that are predictive of disease onset. In this
study we investigated this ideas in a real example, the insulin-producing
pancreatic islet ?-cells and the onset of type 1 diabetes. Within each islet, the ?-cells are
electrically coupled to each other. This intercellular coupling enables
the ?-cells to synchronize their insulin release, thereby generating the multi-scale
temporal rhythms in blood insulin that are critical to maintaining blood glucose
homeostasis. Using percolation theory we show how normal islet function is
intrinsically linked to network connectivity. Percolation is a geometric phase transition of
network structure that has profound impact to the function and dynamics of the
network. We found that the critical amount of ?-cell death at which the islet cellular
network loses site percolation, is consistent with laboratory and clinical
observations of the threshold loss of ?-cells that causes islet functional failure. In
addition, numerical simulations confirm that the islet cellular network needs to be
percolated for ?-cells to synchronize. Furthermore, the interplay between site
percolation and bond strength predicts the existence of a transient phase of islet functional
recovery after onset and introduction of treatment, potentially explaining
a long time mystery in the clinical study of type 1 diabetes: the honeymoon
phenomenon. Based on these results, we hypothesized that the onset of T1D may be the
result of a phase transition of the islet ?-cell network. We will further discuss the
potential applications, as well as the importance of such approaches in the study of
complex networks in living systems.

Date, Location: 
2014-05-01

Professor Martha Alibabli 4/25/2014

Date: 4/25/2014
Time: 3:30PM-4:30PM
Place: 422 Armstrong Hall

Martha Alibabli

Abstract:
Teachers use a range of modalities to communicate in the classroom. In this talk, I focus on how teachers use gestures during instructional communication in mathematics. The bulk of the data are drawn from a corpus of middle school mathematics lessons covering a range of topics. I focus specifically on the role of teachers’ gestures in their communication in two discourse contexts: (1) segments of the lessons in which teachers attempt to connect ideas, concepts, or procedures, and (2) teachers’ responses to "trouble spots" in the classroom discourse (i.e., points where students do not understand or misunderstand the instructional material). The data reveal that gesture is an integral part of classroom communication, and that teachers adjust their gestures depending on lesson content and students’ needs.

Date, Location: 
2014-04-25

Professor Hao Li 4/24/2014

Half of grid graphs are
non-singular

Date: 4/24/2014
Time: 3:30PM-4:30PM
Place: 315 Armstrong Hall

Hao Li

Abstract: Here
TEX: Here

Date, Location: 
2014-04-24

Professor Emanuel Indrei 4/24/2014

Date: 4/24/2014
Time: 3:30PM-4:30PM
Place: 313 Armstrong Hall

Emanuel Indrei

Abstract: Obstacle-type problems appear in various branches of minimal
surface theory, potential theory, and optimal control. In this talk, we
discuss the optimal regularity of solutions to fully nonlinear
obstacle-type free boundary problems. This represents joint work with
Andreas Minne.

Date, Location: 
2014-04-24

Pages

Subscribe to