This paper presents a new algorithm for a batch of task makespan minimisation in heterogeneous multigrid computing. Heterogeneous grids are known to cause straggling task problem that increases task execution makespan. Existing task distribution algorithms solve this problem by using information about the compute node capacities or task sizes. However, such information may not always be available. Task stalling solves both problems. However, this method is described for queuing systems consisting of only two heterogeneous servers or grids. Our proposed algorithm is based on an improved task stalling method, allowing it to distribute tasks in systems consisting of two or more grids. Experiment results show reduced task execution makespan by up to 19,92% compared to FIFO. This allows us to conclude that the new algorithm is suitable for a batch of task makespan minimisation in heterogeneous multigrid computing.
Section 1 of this paper follows entirely a scenario from the article “Engineering Compliant Software: Advising Developers by Automating Legal Reasoning” by D. Oberle, F. Drefs, R. Wacker, C. Baumann and O. Raabe, SCRIPTed (2012) 9:3, 280–313, where it serves as a running example. It demonstrates that data transfer violates the law. This motivating scenario has added value in the education of software developers and is worth sharing with the computer communities of other countries including Lithuania. In the scenario, the continental law and EU law sways the particularities of the German law. The motivation for teaching the scenario can be compared with teaching concrete cases in the study of law. Legal reasoning is demonstrated by supplementing the provisions of the German Federal Data Protection Act (FDPA) with those of the Lithuanian Law on Legal Protection of Personal Data, which have the same meaning. In Section 3, we attempt to formulate the software compliance problem. Finally, we explain the notion of subsumption – a legal qualification of facts according to a norm’s circumstance. We consider subsumption to consist of two notions: terminological subsumption and normative subsumption.
Bayesian Networks are used to model a user's behaviour. There is not much research on the use of Frequentist Inference to accomplish this same task. This paper aims to analyze and describe the differences between inference methods: Bayesian and Frequentist. A simulation was conducted using Conditional Probabilities that were drawn from the Drupal Usability Study that was conducted in 2012 to apply to both inference methods, Bayesian and Frequentist. Results from this simulation showed that for most probabilities, Bayesian and Frequentist values are reasonably close. Although more frequentist values were equal to 50% than Bayesian values. With this, it was deduced that for Adaptive User Interfaces, Bayesian Inference is a superior method to use.
When doing a searching process, Binary Search is one of the classic algorithm used in sorted data. The characteristic of this algorithm is to make a comparison of the keywords you want to find with the start, middle, and end values of a data series. Keyword search is done by reducing the range of start and end points to finally find the keyword you want to search. The time complexity of the binary search algorithm is O(log2n) while the memory capacity needed is O(1) for iterative implementation and O(log2n) for recursive implementation. This research will develop a level of comparison in binary search in order to get optimal performance in accordance with the amount of data available.
A method for the calculation of the one-particle generalized coefficients of fractional parentage for an arbitrary number of j-orbits with isospin and an arbitrary number of oscillator quanta (generalized CFPs or GCFPs) is presented. The approach is based on a simple enumeration scheme for antisymmetric many-particle states, an efficient algorithm for the calculation of the CFPs for a single j-orbit with isospin, and a general procedure for the computation of the angular momentum (isospin) coupling coefficients describing the transformation between different momentum-coupling schemes. The method provides fast calculation of GCFPs for a given particle number and produces results possessing small numerical uncertainties. The introduced GCFPs make it feasible calculation of expectation values of one-particle nuclear shell-model operators within the isospin formalism.
This paper is focused on the Bayes approach to multiextremal optimization problems, based on modelling the objective function by Gaussian random field (GRF) and using the Euclidean distance matrices with fractional degrees for presenting GRF covariances. A recursive optimization algorithm has been developed aimed at maximizing the expected improvement of the objective function at each step, using the results of the optimization steps already performed. Conditional mean and conditional variance expressions, derived by modelling GRF with covariances expressed by fractional Euclidean distance matrices, are used to calculate the expected improvement in the objective function. The efficiency of the developed algorithm was investigated by computer modelling, solving the test tasks, and comparing the developed algorithm with the known heuristic multi-extremal optimization algorithms.
In this paper, we are analyzing the results of native Lithuanian speaker recognition and identification using long short-term memory deep neural network. We look at recognition accuracy and identify further potential improvements. Dataset used for training and speaker recognition consists of over 370 unique speakers, who provide their voice utterances in Lithuanian language. In this paper we present results that are derived from part of this dataset.
Population initialization is one of the important tasks in evolutionary and genetic algorithms (GAs). It can affect considerably the speed of convergence and the quality of the obtained results. In this paper, some heuristic strategies (procedures) for construction of the initial populations in genetic algorithms are investigated. The purpose is to try to see how the different population initialization strategies (procedures) can influence the quality of the final solutions of GAs. Several simple procedures were algorithmically implemented and tested on one of the hard combinatorial optimization problems, the quadratic assignment problem (QAP). The results of the computational experiments demonstrate the usefulness of the proposed strategies. In addition, these strategies are of quite general character and may be easily transferred to other population-based metaheuristics (like particle swarm or bee colony optimization methods).
Occurrence of the agent paradigm and its further applications have stimulated the emergence of new concepts and methodologies in computer science. Today terms like multi-agent system, agent-oriented methodology, and agent-oriented programming (AOP) are widely used. The aim of this paper is to clarify the validity of usage of the terms AOP and AOP language. This is disclosed in two phases of an analysis process. Determining to which concepts, terms like agent, programming, object-oriented analysis and design, object-oriented programming, and agent-oriented analysis and design correspond is accomplished in the first phase. Analysis of several known agent system engineering methodologies in terms of key concepts used, final resulting artifacts, and their relationship with known programming paradigms and modern tools for agent system development is performed in the second phase. The research shows that in the final phase of agent system design and in the coding stage, the main artifact is an object, defined according to the rules of the object-oriented paradigm. Hence, we conclude that the computing society still does not have AOP owing to the lack of an AOP language. Thus, the term AOP is very often incorrectly assigned to agent system development frameworks that in all cases, transform agents into objects.
Šiame darbe sudarytas rekurentinis paslėptųjų Markovo modelių parametrų vertinimo algoritmas. Paslėptieji Markovo modeliai modeliuojami Gauso skirstiniu, kurio parametrai pasiskirstę pagal daugiamatį normalųjį dėsnį su nežinomais vidurkių vektoriumi ir kovariacijų matrica. Nežinomų parametrų įverčiai gaunami didžiausio tikėtinumo metodu. Rekurentinis algoritmas sudarytas remiantis didžiausio tikėtinumo metodu išvestomis formulėmis ir klasikiniu EM algoritmu. Kadangi rekurentinio algoritmo vykdymo laikas yra proporcingas apdorojamų stebėjimų skaičiui, tai jis gali būti naudojamas modelio parametrų vertinimui realiu laiku. Realizuoto rekurentinio EM algoritmo savybės buvo ištirtos kompiuteriniu eksperimentu klasterizuojant duomenis. Jis taip pat gali būti taikomas duomenų klasifikavimo ir atpažinimo realiu laiku uždaviniams spręsti.