Academic research is driven by several factors causing different disciplines to act as “sources” or “sinks” of knowledge. However, how the flow of authors’ research interests – a proxy of human knowledge – evolved across time is still poorly understood. Here, we build a comprehensive map of such flows across one century, revealing fundamental periods in the raise of interest in areas of human knowledge. We identify and quantify the most attractive topics over time, when a relatively significant number of researchers moved from their original area to another one, causing what we call a “diaspora of the knowledge” towards sinks of scientific interest, and we relate these points to crucial historical and political events. Noticeably, only a few areas – like Medicine, Physics or Chemistry – mainly act as sources of the diaspora, whereas areas like Material Science, Chemical Engineering, Neuroscience, Immunology and Microbiology or Environmental Science behave like sinks.
Quantifying the diaspora of knowledge in the last century
Manlio De Domenico, Elisa Omodei and Alex Arenas
Applied Network Science20161:15
Similar to the Autonomous Computing initiative, which has mainly been advancing techniques for self-optimization focusing on computing systems and infrastructures, Organic Computing (OC) has been driving the development of system design concepts and algorithms for self-adaptive systems at large. Examples of application domains include, for instance, traffic management and control, cloud services, communication protocols, and robotic systems. Such an OC system typically consists of a potentially large set of autonomous and self-managed entities, where each entity acts with a local decision horizon. By means of cooperation of the individual entities, the behavior of the entire ensemble system is derived. In this article, we present our work on how autonomous, adaptive robot ensembles can benefit from OC technology. Our elaborations are aligned with the different layers of an observer/controller framework, which provides the foundation for the individuals’ adaptivity at system design-level. Relying on an extended Learning Classifier System (XCS) in combination with adequate simulation techniques, this basic system design empowers robot individuals to improve their individual and collaborative performances, e.g., by means of adapting to changing goals and conditions. Not only for the sake of generalizability but also because of its enormous transformative potential, we stage our research in the domain of robot ensembles that are typically comprised of several quad-rotors and that organize themselves to fulfill spatial tasks such as maintenance of building facades or the collaborative search for mobile targets. Our elaborations detail the architectural concept, provide examples of individual self-optimization as well as of the optimization of collaborative efforts, and we show how the user can control the ensembles at multiple levels of abstraction. We conclude with a summary of our approach and an outlook on possible future steps.
An Organic Computing Approach to Self-Organizing Robot Ensembles
Sebastian von Mammen, Sven Tomforde and Jörg Hähner
Front. Robot. AI, 17 November 2016 | https://doi.org/10.3389/frobt.2016.00067
How much information do large brain networks integrate as a whole over the sum of their parts? Can the dynamical complexity of such networks be globally quantified in an information-theoretic way and be meaningfully coupled to brain function? Recently, measures of dynamical complexity such as integrated information have been proposed. However, problems related to the normalization and Bell number of partitions associated to these measures make these approaches computationally infeasible for large-scale brain networks. Our goal in this work is to address this problem. Our formulation of network integrated information is based on the Kullback-Leibler divergence between the multivariate distribution on the set of network states versus the corresponding factorized distribution over its parts. We find that implementing the maximum information partition optimizes computations. These methods are well-suited for large networks with linear stochastic dynamics. We compute the integrated information for both, the system’s attractor states, as well as non-stationary dynamical states of the network. We then apply this formalism to brain networks to compute the integrated information for the human brain’s connectome. Compared to a randomly re-wired network, we find that the specific topology of the brain generates greater information complexity.
The global dynamical complexity of the human brain network
Xerxes D. Arsiwalla and Paul F. M. J. Verschure
Multistable systems exhibit a rich front dynamics between equilibria. In one-dimensional scalar gradient systems, the spread of the fronts is proportional to the energy difference between equilibria. Fronts spreading proportionally to the energetic difference between equilibria is a characteristic of one-dimensional scalar gradient systems. Based on a simple nonvariational bistable model, we show analytically and numerically that the direction and speed of front propagation is led by nonvariational dynamics. We provide experimental evidence of nonvariational front propagation between different molecular orientations in a quasi-one-dimensional liquid-crystal light valve subjected to optical feedback. Free diffraction length allows us to control the variational or nonvariational nature of this system. Numerical simulations of the phenomenological model have quite good agreement with experimental observations.
Nonvariational mechanism of front propagation: Theory and experiments
A. J. Alvarez-Socorro, M. G. Clerc, G. González-Cortés, and M. Wilson
Phys. Rev. E 95, 010202(R) – Published 17 January 2017