A characteristic feature of many practical decision-making tasks is their multicriteria, which leads to the complexity of information processing when finding a solution. Failure to consider many criteria can lead to i...
详细信息
With the rapid advancements of sensor technologies and mobile computing, Mobile Crowd-Sensing (MCS) has emerged as a new paradigm to collect massive-scale rich trajectory data. Nomadic sensors empower people and objec...
详细信息
ISBN:
(数字)9783030380816
ISBN:
(纸本)9783030380816;9783030380809
With the rapid advancements of sensor technologies and mobile computing, Mobile Crowd-Sensing (MCS) has emerged as a new paradigm to collect massive-scale rich trajectory data. Nomadic sensors empower people and objects with the capability of reporting and sharing observations on their state, their behavior and/or their surrounding environments. Processing and analyzing this continuously growing data raise several challenges due not only to their volume, their velocity, and their complexity but also to the gap between raw data samples and the desired application view in terms of correlation between observations and in terms of granularity. In this paper, we put forward a proposal that offers an abstract view of any spatio-temporal data series as well as their manipulation. Our approach allows to support this high-level logical view and provides efficient processing by mapping both the representation and the manipulation to an internal physical model. We explore an implementation within a distributed framework and envision the adaptation of data organization methods combining aggressive indexing and partitioning over time and space. The mapping from the logical view and the actual data storage will lead to revisiting the traditional database query rewriting and optimization techniques. This proposal is a first step in the objective of coping with the complexity, the imperfection of large data sizes in the MCS context.
The outcome of a machine learning algorithm is a prediction model. Typically, these models are computationally expensive, where improving of the quality the prediction leads to a decrease in the inference speed. Howev...
详细信息
The outcome of a machine learning algorithm is a prediction model. Typically, these models are computationally expensive, where improving of the quality the prediction leads to a decrease in the inference speed. However it is not always tradeoff between quality and speed. In this paper we show it is possible to speed up the model by using additional memory without losing significat prediction quality for a novel boosted trees algorithm called CatBoost. The idea is to combine two approaches: training fewer trees and merging trees into a kind of hashmaps called DecisionTensors. The proposed method allows for pareto-optimal reduction of the computational complexity of the decision tree model with regard to the quality of the model. In the considered example the number of lookups was decreased from 5000 to only 6 (speedup factor of 1000) while AUC score of the model was reduced by less than 10(-3).
The bright future of particle physics at the Energy and Intensity frontiers poses exciting challenges to the scientific software community. The traditional strategies for processing and analysing data are evolving in ...
详细信息
The bright future of particle physics at the Energy and Intensity frontiers poses exciting challenges to the scientific software community. The traditional strategies for processing and analysing data are evolving in order to (i) offer higher-level programming model approaches and (ii) exploit parallelism to cope with the ever increasing complexity and size of the datasets. This contribution describes how the ROOT framework, a cornerstone of software stacks dedicated to particle physics, is preparing to provide adequate solutions for the analysis of large amount of scientific data on parallel architectures. The functional approach to parallel dataanalysis provided with the ROOT TdataFrame interface is then characterised. The design choices behind this new interface are described also comparing with other widely adopted tools such as Pandas and Apache Spark. The programming model is illustrated highlighting the reduction of boilerplate code, composability of the actions and data transformations as well as the capabilities of dealing with different data sources such as ROOT, JSON, CSV or databases. Details are given about how the functional approach allows transparent implicit parallelisation of the chain of operations specified by the user. The progress done in the field of distributed analysis is examined. In particular, the power of the integration of ROOT with Apache Spark via the PyROOT interface is shown. In addition, the building blocks for the expression of parallelism in ROOT are briefly characterised together with the structural changes applied in the building and testing infrastructure which were necessary to put them in production.
In this paper, security of secret key extraction scheme is evaluated for private communication between legitimate wireless devices. Multiple adversaries that distribute around these legitimate wireless devices eavesdr...
详细信息
ISBN:
(纸本)9781538643280
In this paper, security of secret key extraction scheme is evaluated for private communication between legitimate wireless devices. Multiple adversaries that distribute around these legitimate wireless devices eavesdrop on the data transmitted between them, and deduce the secret key. Conditional min-entropy given the view of those adversaries is utilized as security evaluation metric in this paper. Besides, the wiretap channel model and hidden Markov model (HMM) are regarded as the channel model and a dynamic programming approach is used to approximate conditional min-entropy. Two algorithms are proposed to mathematically calculate the conditional minentropy by combining the Viterbi algorithm with the Forward algorithm. Optimal method with multiple adversaries (OME) algorithm is proposed firstly, which has superior performance but exponential computation complexity. To reduce this complexity, suboptimal method with multiple adversaries (SOME) algorithm is proposed, using performance degradation for the computation complexityreduction. In addition to the theoretical analysis, simulation results further show that the OME algorithm indeed has superior performance as well as the SOME algorithm has more efficient computation.
The automotive industry seeks to include more and more features in its vehicles. For this purpose, the necessary policy shift towards multi-core technology is in full swing. To eventually exploit the extra processing ...
详细信息
The automotive industry seeks to include more and more features in its vehicles. For this purpose, the necessary policy shift towards multi-core technology is in full swing. To eventually exploit the extra processing power, there is much additional effort needed for coping with the tremendously increased complexity. This is largely due to the elaborate parallelization process that spans a vast search space. Consequently, there is a strong need for innovative methods and appropriate tools for the migration of legacy single-core software. We use the results of a data dependency analysis performed on AUTOSAR system descriptions to determine advantageous partitions as well as initial task-to-core mappings. Afterwards, the extracted information serves as input for the simulation within a multi-core timing tool suite. Here, the initial solution is evaluated with respect to proper scheduling and metrics like cross-core communication rates, communication latencies, or core load distribution. A subsequent optimization process improves the initial solution and enables a comparative assessment. To demonstrate the benefit, we substantially expand a previous case study by applying our approach to two complex engine management systems and by showing the advantages compared to a parallelization process without preceding dependency analysis and initial partition/mapping suggestions.
To help produce accurate and consistent maritime hazard products, the National Tsunami Hazard Mitigation Program organized a benchmarking workshop to evaluate the numerical modeling of tsunami currents. Thirteen teams...
详细信息
To help produce accurate and consistent maritime hazard products, the National Tsunami Hazard Mitigation Program organized a benchmarking workshop to evaluate the numerical modeling of tsunami currents. Thirteen teams of internationalresearchers, using a set of tsunami models currently utilized for hazard mitigation studies, presented results for a series of benchmarking problems;these results are summarized in this paper. Comparisons focus on physical situations where the currents are shear and separation driven, and are thus de-coupled from the incident tsunami waveform. In general, we find that models of increasing physical complexity provide better accuracy, and that low-order three-dimensional models are superior to high-order two-dimensional models. Inside separation zones and in areas strongly affected by eddies, the magnitude of both model-data errors and inter-model differences can be the same as the magnitude of the mean flow. Thus, we make arguments for the need of an ensemble modeling approach for areas affected by large-scale turbulent eddies, where deterministic simulation may be misleading. As a result of the analyses presented herein, we expect that tsunami modelers now have a better awareness of their ability to accurately capture the physics of tsunami currents, and therefore a better understanding of how to use these simulation tools for hazard assessment and mitigation efforts. (C) 2017 Elsevier Ltd. All rights reserved.
The management of uncertainty is a critical aspect of current as well as future air traffic control operations. This study investigated: (1) sources of uncertainty in enroute air traffic control, (2) strategies that a...
详细信息
The management of uncertainty is a critical aspect of current as well as future air traffic control operations. This study investigated: (1) sources of uncertainty in enroute air traffic control, (2) strategies that air traffic controllers adopt to cope with uncertainty, (3) the trade-offs and contingencies that influences the adoption of these uncertainties, and (4) the requirements for system design that support controllers in following these strategies. The data were collected using a field study in two enroute air traffic control centres, involving "over the shoulder" observation sessions, discussions with air traffic controllers, and document analysis. Three types of uncertainty coping strategies were identified: reducing uncertainty, acknowledging uncertainty, and increasing uncertainty. The RAWFS heuristic (Lipshitz and Strauss in Organ Behav Hum Decis Process 69:149-163, 1997) and anticipatory thinking (Klein et al. in Anticipatory thinking, Proceedings of the eighth international NDM conference, Pacific Grove, CA, 2007) were used to identify reduction and acknowledgement strategies. Recent suggestions by Grote (Saf Sci 71:71-79, 2015) were used to further explore strategies that increase uncertainty. The study presents a new framework for the classification of uncertainties in enroute air traffic control and identified the uncertainty management strategies and underlying tactics, in context of contingencies and trade-offs between operational goals. The results showed that controllers, in addition to reducing and acknowledging uncertainty, may deliberately increase uncertainty in order to increase flexibility for other actors in the system to meet their operational goals. The study describes new tactics for acknowledging and increasing uncertainty. The findings were summarized in the air traffic controller complexity and uncertainty management model. Additionally, the results bring to light system design recommendations that allow controllers to follow these
This paper concerns the dimension reduction in regression for large data set. The authors introduce a new method based on the sliced inverse regression approach, cMled cluster-based regularized sliced inverse regressi...
详细信息
This paper concerns the dimension reduction in regression for large data set. The authors introduce a new method based on the sliced inverse regression approach, cMled cluster-based regularized sliced inverse regression. The proposed method not only keeps the merit of considering both response and predictors＇ information, but also enhances the capability of handling highly correlated variables. It is justified under certain linearity conditions. An empirical application on a macroeconomic data set shows that the proposed method has outperformed the dynamic factor model and other shrinkage methods.
The proceedings contain 17 papers. The topics discussed include: the use of global sensitivity methods for the analysis, evaluation and improvement of complex modeling systems;optimization and linear control of large ...
ISBN:
(纸本)9783642149405
The proceedings contain 17 papers. The topics discussed include: the use of global sensitivity methods for the analysis, evaluation and improvement of complex modeling systems;optimization and linear control of large scale nonlinear systems: a review and a suite of modelreduction-based techniques;universal algorithms, mathematics of semirings and parallel computations;scaling invariant interpolation for singularly perturbed vector fields (SPVF);think globally, move locally: coarse graining of effective free energy surfaces;extracting functional dependence from sparse data using dimensionality reduction: application to potential energy surface construction;a multilevel algorithm to compute steady states of lattice Boltzmann models;time step expansions and the invariant manifold approach to lattice Boltzmann models;and adaptive simplification of complex systems: a review of the relaxation- redistribution approach.
暂无评论