Past CNS Talks
Robustness, clustering and evolutionary conservation of protein-protein interactions
Abstract: Contemporary genomics and proteomics tries to elucidate the webs of protein-protein interaction networks of various organisms. A critical goal of the network paradigm thus emerging is the potential for application of these principles to understand and make predictions about biological systems. Yet, the severe error proneness of methods to elucidate protein-protein interactions casts doubt on the general usefulness of the resulting topologies. Modeling the high error rates in the determination of both protein-protein interactions and orthologous proteins in yeast, we find that the originally reported trends on the preferential evolutionary conservation of highly interacting proteins and cohesive groups of proteins are robust.
The Expression and Control of Attentional Topography in Human Visual Cortex
Abstract: Using fMRI it has been possible to study how attention spatially modulates visual processing in the occipital lobe under various stimulus and task conditions. However, this does not account for how other areas, such as fronto-parietal cortex, may control these attentional effects. In a series of experiments, we examined the topographic representation of attended locations in both occipital and parietal cortex in order to understand better how the expression of spatial attention and its control are interrelated.
Abstract: This talk will describe ongoing efforts to study the topological and dynamical properties of link, lexical, and semantic networks stemming from various features of the World Wide Web. I will outline what we think, what we know, what we can use regarding the structure and content of the Web, and what the future of intelligent Web search may bring.
Computer-Mediated Social Networks
Abstract: The near ubiquitous use of computer media has stirred controversy about what is appropriate, possible, and efficient to do via these media. While many studies have looked at what use is made of CMC, studies have mainly examine aggregate views of single media, for example, looking at how email, bulletin boards, or blogs are used. This ignores the very real role of interpersonal ties on media use and relational maintenance, and the way multiple media support ties and group interactions.
Unraveling the Biochemical Reaction Kinetics from Time-Series Data
Abstract: Time course data can now be routinely collected for biochemical reaction pathways, and recently, we are proposing several methods to infer reaction mechanisms for metabolic pathways and networks. In this talk we provide a survey of techniques for determining reaction mechanisms for time course on the concentration or abundance of different reacting components, with little prior information about the pathways involved.
Understanding Community Ecology through Network Analysis
J. Alison Bryant
Abstract: Recent work in organizational change has highlighted three opportunities for future investigation - the need to understand organizational evolution from the level of the community; the need to more systematically understand the complex relationships within the community; and the need to incorporate network analysis in the study of community ecology.
Statistical Physics of Popularity-Driven Networks
Abstract: The rate equation approach is applied to quantify basic features of growing, popularity-driven networks. A prototypical example is the network of citations associated with scientific publications. Basic empirical facts about the citation network will be presented, based on the entire corpus of Physical Review publications from the past 110 years.
Organization, Development, and Function of Complex Brain Networks
Abstract: Recent research has revealed general principles in the structural and functional organization of complex networks which are shared by various natural, social and technological systems. This talk examines these principles as applied to the organization, development and function of complex brain networks.
Evolving Neural Network Architectures in a Computational Ecology
Abstract: Profound evidence exists to demonstrate wide-spread, general plasticity and learning in biological brains, yet equally clearly the "wiring diagram" of the brain matters. Key attributes of brain function and form have been shown to be well modeled by Hebbian learning in artificial neural networks (ANN's) with suitable network architectures. I will discuss an artificial life system designed to evolve highly arbitrary ANN architectures, which then employ Hebbian learning, in a computational ecology.
Groups as Crucial Connectors in Networks
Abstract: This paper shows how basic properties of inequality in directed networks have systematic implications for connectedness in networks. We specifically show how four particular properties of inequality in directed networks are related to the frequencies of various types of indirect connections.
Emergent Properties in Local Networks of Cortical Neurons
John M. Beggs
Abstract: The cerebral cortex has expanded rapidly in the evolution of mammals and is essential for higher cognition. Despite a surface area of nearly 2500 cm2, human cortex remarkably seems to be composed of many similar modules, each ~0.5 cm2 and containing about 150,000 neurons. Each module receives information from other modules, performs some operations, and passes the results on to other modules. But how do these modules themselves process information?
Coordination in Social-Ecological Networks: The Case of Irrigation on Bali
Abstract: Various local and regional social-ecological systems (SES's) have existed for hundreds of years, remaining in particular configurations that have withstood a variety of natural and social disturbances. What enabled these systems to persist? Many long-lived SES's have adapted their institutions to the disturbance and stress regime they have experienced over time as well as to the broader economic, political, and social system in which they are located. Such adaptations change the use of resources in time and/or space to maintain the desired configuration of the SES's.
Epidemic Modeling: Dealing with Complexity
Abstract: The mathematical modeling of epidemics is a very active field of research that crosses different disciplines. Epidemiologists, computer scientists and social scientists share a common interest in studying spreading phenomena and make use of very similar models for the description of the diffusion of viruses, knowledge and innovation. Epidemic modeling relies also on the knowledge of the underlying population structure in which the spreading is occurring. In this perspective, the increased power of computers and informatics tools is having a large impact on epidemic modeling by allowing the gathering and handling of large data sets for a variety of contact networks of practical interest in social science, critical infrastructures and epidemiology.
Why Should I Care about Social Network Enterprise Software?
Abstract: This talk, which could be subtitled "What I Did on my Summer Vacation", describes scientific work at VisiblePath Corporation, in New York City.
VisiblePath's approach to social network analysis in the enterprise is based on four core tenets (taken directly from http://www.visiblepath.com):
1. Social networks are complex. Typical corporate networks are millions of times as complex as a simple social network structure. Complexity is a function of robust attribute data on the nodes (people), multiple relations and links between nodes, valued relations, degradation of relations over time, and the breadth and density of enterprise networks, where typical networks include tens of millions of relations between millions of nodes.
2. Supporting data structures are complex. Simple networks can be rendered as sociomatrices. Complex networks required tiered sociomatrices for intelligent analysis and application of SNA theory. Complexity is compounded by the disparate sources of relational data which include Customer Relations Management (CRM) systems, messaging applications, desktop data, other enterprise sources, and the lack of common structure in the associated data models.
3. Data-oriented methodology drives efficient Relationship Capital Management (RCM). Intelligent analysis has a significant impact on the value of an RCM system. At later stages of deployment, efficient network path analysis delivers higher close rates and shorter sales cycles. At early stages of deployment, efficient analysis dramatically reduces spam in the system that can undercut broad enterprise adoption.
4. RCM analytics power 3rd party applications. The Visible Path platform is designed to enhance and extend existing enterprise applications. The value of the platform is derived from proprietary modules that have been developed by Visible Path to manage the quantity and complexity of data required to deliver effective application level functionality.
This talk focuses on the network analysis aspects of VisiblePath's software.