I recently attended Shreyas Sundaram's talk at the University of Waterloo. Shreyas is an assistant professor in electrical and computer engineering. Some aspects of the talk are more technical than a general audience will likely tolerate but it is worth sticking with Shreyas through those parts for the conclusions and insights that his work represents.
A couple of points to note - the ability to discover malicious nodes mathematically in a network. This is useful in finding bad sensors but is equally useful in understanding where key gatekeepers are in a human network. The brokers who are critical to information flows have always been important but in vary large data sets they may be difficult to detect.
Another interesting point very much related to the malicious nodes work is the ability to determine which nodes bridge between network structures. What nodes, if you removed them, would split the network into two or more distinct parts? There are many possible applications of this as well.
The power of developing powerful mathematical tools for understanding networks is that the processes developed can be applied to very large data sets. Mathematics also makes it possible to see novel structures and qualities that would not be possible with human processing alone. From epidemics to information technology structures and power grids, gaining greater insight into the science of networks is critical. In addition, disseminating that information and teaching people across disciplines how the technical aspects of networks function will determine the level of sophistication we can sustain in our various domains of analysis.
The Waterloo Institute for Complexity and Innovation was the host of the talk and they seek to deepen our understanding and application of complexity science as a means of approaching the most challenging issues and questions of our time.
Diffusing Information and Reaching Agreement in Networks: Convergence and Resilience from Waterloo Institute for Complexit on Vimeo.
(that's me ducking in at the front right at the beginning)
A couple of points to note - the ability to discover malicious nodes mathematically in a network. This is useful in finding bad sensors but is equally useful in understanding where key gatekeepers are in a human network. The brokers who are critical to information flows have always been important but in vary large data sets they may be difficult to detect.
Another interesting point very much related to the malicious nodes work is the ability to determine which nodes bridge between network structures. What nodes, if you removed them, would split the network into two or more distinct parts? There are many possible applications of this as well.
The power of developing powerful mathematical tools for understanding networks is that the processes developed can be applied to very large data sets. Mathematics also makes it possible to see novel structures and qualities that would not be possible with human processing alone. From epidemics to information technology structures and power grids, gaining greater insight into the science of networks is critical. In addition, disseminating that information and teaching people across disciplines how the technical aspects of networks function will determine the level of sophistication we can sustain in our various domains of analysis.
The Waterloo Institute for Complexity and Innovation was the host of the talk and they seek to deepen our understanding and application of complexity science as a means of approaching the most challenging issues and questions of our time.
Diffusing Information and Reaching Agreement in Networks: Convergence and Resilience from Waterloo Institute for Complexit on Vimeo.
(that's me ducking in at the front right at the beginning)
No comments:
Post a Comment