Session #1 Abstracts

Playing Hard to Get: a Counterintuitive Means to Promote Parasitism

Roy Wright

Parasites and other exploiters of herbivore populations can provide tremendous benefits to plant life by protecting it from destruction, in a process known as a trophic cascade. Thus, the preservation of certain parasites, such as entomopathogenic nematodes, can be a potent form of natural pest control both in the wild and in agricultural situations. Past mathematical modeling of one such nematode-host interaction showed that the parasitic population should undergo violent two-year cycles, with dangerously low numbers every other year. However, field observations indicate that the interaction may often be fairly persistent. Building upon the previous model, we investigate the possible effects of an alternate host on nematode persistence, and find that the persistence time of the parasitic nematode is extended only if the second host is sufficiently inaccessible.

Fully-Automated White Matter Hyperintensity Detection With Anatomical Prior Knowledge and Without FLAIR

Christopher Schwarz, Evan Fletcher, Charles DeCarli, Owen Carmichael

White Matter Hyperintensities (WMHs) are dysfunctional regions of the brain whose occurrence strongly correlates to the presence and progression of several neurological conditions such as Alzheimer's disease. This work presents a method that employs a Markov Random Field (MRF) approach for detection of these WMHs based on run-time PD-, T1-, and T2-weighted structural magnetic resonance (MR) images of the brain along with labeled training examples. Unlike most prior approaches, the method is able to reliably detect WMHs in the absence of fluid-attenuated (FLAIR) images. Its success is due to the learning of probabilistic models of WMH spatial distribution and neighborhood dependencies from ground-truth examples of FLAIR-based WMH detections. These models are combined with a probabilistic lognormal mixture model of the PD, T1, and T2 intensities of WMHs in a MRF framework that provides the machinery for inferring the positions of WMHs in novel test images. The method is shown to accurately detect WMHs in a set of 126 elderly subjects from an academic dementia clinic. The use of anatomical prior knowledge that captures the known asymmetric anatomical course of WMH progression is shown to increase the accuracy of WMH detection over MRFs that smooth the WMH detections isotropically.

Topological Influences on Function in a Spiking Neural Network

Richard Watson, Nick Travers

Networks, and neural networks in particular, are a rapidly expanding field of study, with many promising applications. However, the relation between topology and function in neural networks is still poorly understood. Using two measures of functional similarity and exploring three topological classes of connectivity, we have examined the extent to which a spiking neural network.s function changes as its topology is modified. We find that the degree to which functionality is altered depends on both network size and the underlying base topology of the original reference network. In general, smaller networks tend to show less average functional change for a given number of edge modifications, but also much larger variability in the amount of this change. Additionally, networks generated using a small- world model [1] display significantly less functional change than networks with scale-free degree distributions or simple Erdos-Renyi random graphs.

An Evaluation of Identity-Sharing Behavior, Privacy Concerns and Trust in a Social Network Community

Avinash Nayak, Raissa D'Souza

It is not well understood how individual privacy concerns and trust influence social interactions within social networking sites. A survey of a popular social networking site, Facebook, compared perceptions of trust and privacy concerns, along with the willingness to share information to develop new relationships. The survey data reported different levels of privacy concern and the use of the Facebook account. Facebook members expressed significantly greater trust in Facebook and in other Facebook members, and hence were more willing to share identifying information even though many Facebook users (in this study) have not read the .Facebook.s Privacy Policy. but still feel that their privacy is well protected by Facebook. Many in this study use Facebook more than e-mail as the .new medium. to communicate. These results suggest that in online interaction, trust is not as necessary in building new relationships unlike face to face encounters. This study also shows that in a social networking site, trust and the willingness to share information do not automatically translate into new social interaction. This study demonstrates the direct relationship that online interaction has with the amount of time spent on that site.

P, NP and the Nullstellensatz: Independent Set and Graph-3-Coloring Infeasibility Certificates

Susan Margulies, Jesus De Loera, Jon Lee, Peter Malkin

Systems of polynomial equations over an algebraically-closed field K can be concisely used to model many combinatorial problems. In this way, a combinatorial problem is feasible (e.g., a graph is 3-colorable or has an independent set of size k) if and only if a related system of polynomial equations has a solution over K. If the combinatorial problem is infeasible, Hilbert's Nullstellensatz and a large-scale linear algebra computation yields a certificate of infeasibility. Thus, unless P = NP, there must exist an infinite sequence of infeasible instances for each hard combinatorial problem where the minimum-degree of a Hilbert Nullstellensatz infeasibility certificate grows.
We show that the minimum-degree of a Nullstellensatz certificate for the non-existence of an independent set of size greater than the size of the largest independent set in the graph is the size of the largest independent set in the graph. Moreover, such a certificate contains at least one term per independent set in G. By contrast, for graph-3-colorability, the Nullstellensatz-Linear Algebra (NulLA) algorithm proves the infeasibility of instances having thousands of nodes and tens of thousands of edges.

Back to conference program
Back to conference main page