35 research outputs found
Enhancing Command Communications and Innovation with SAINT:Semantics, Adaptation, and Influence in Networked Teams
No full textKnowledge-Independent Data Mining with Fine-Grained Parallel Evolutionary Algorithms
No full textThis paper illustrates the application of evolutionary algorithms (EA) to data mining problems. The objectives are to demonstrate that EA can provide a competitive general purpose data mining scheme for classification tasks without constraining the knowledge representation, and that it can be achieved reducing the amount of time required using the inherent parallel processing nature of EA
Inducing Partially-Defined Instances with Evolutionary Algorithms
No full textThis paper addresses the issue of reducing the storage requirements on Instance-Based Learning algorithms. Algorithms proposed by other researches use heuristics to prune instances of the training set or modify the instances themselves to achieve a reduced set of instances. Our work presents an alternative way. We propose to induce a reduced set of partially-defined instances with Evolutionary Algorithms. Experiments were performed with GALE, our fine-grained parallel Evolutionary Algorithm, and other well-known reduction techniques on several datasets. Results suggest that Evolutionary Algorithms are competitive and robust for inducing sets of partially-defined instances, achieving better reduction rates in storage requirements without losses in generalization accuracy
Optimally designed nanolayered metal-dielectric particles as probes for massively multiplexed and ultrasensitive molecular assays
No full textAn outstanding challenge in biomedical sciences is to devise a palette of molecular probes that can enable simultaneous and quantitative imaging of tens to hundreds of species down to ultralow concentrations. Addressing this need using surface-enhanced Raman scattering-based probes is potentially possible. Here, we theorize a rational design and optimization strategy to obtain reproducible probes using nanospheres with alternating metal and reporter-filled dielectric layers. The isolation of reporter molecules from metal surfaces suppresses chemical enhancement, and consequently signal enhancements are determined by electromagnetic effects alone. This strategy synergistically couples interstitial surface plasmons and permits the use of almost any molecule as a reporter by eliminating the need for surface attachment. Genetic algorithms are employed to optimize the layer dimensions to provide controllable enhancements exceeding 11 orders of magnitude and of single molecule sensitivity for nonresonant and resonant reporters, respectively. The strategy also provides several other opportunities, including a facile route to tuning the response of these structures to be spectrally flat and localization of the enhancement within a specific volume inside or outside the probe. The spectrally uniform enhancement for multiple excitation wavelengths and for different shifts enables generalized probes, wheras enhancement tuning permits a large dynamic range by suppression of enhancements from outside the probe. Combined, these theoretical calculations open the door for a set of reproducible and robust probes with controlled sensitivity for molecular sensing over a concentration range of over 20 orders of magnitude.</jats:p
Mining Directed Social Network from Message Board
No full textIn the paper, we present an approach to mining a directed social network from a message board on the Internet where vertices denote individuals and directed links denote the flow of influence. The influence is measured based on propagating terms among individuals via messages. The distance with respect to contextual similarity between individuals is acquired since the influence indicates the degree of their shared interest represented as terms
The 9th International Workshop on Learning Classifier Systems (IWLCS-2006 Workshop Summary)
No full textA summary of work presented at IWLCS-2006, published in the newsletter of the ACM special interest group on genetic and evolutionary computation.A summary of work presented at IWLCS-2006, published in the newsletter of the ACM special interest group on genetic and evolutionary computation
Evolutionary process indicators for active IGAs applied to weight tuning in unit selection TTS synthesis
No full text
