DOI: 10.3897/jucs.164745

Authors: Aik Tarkhanyan, Ashot Harutyunyan

Abstract: Several studies have shown that the Dempster–Shafer theory (DST) can be successfully applied to scenarios where model interpretability is essential. Although DST-based algorithms offer significant benefits, they face challenges in terms of efficiency. We present a method for the Dempster-Shafer Gradient Descent (DSGD) algorithm that significantly reduces training time—by a factor of 1.6—and also reduces the uncertainty of each rule (a condition on features leading to a class label) by a factor of 2.1, while preserving accuracy comparable to other statistical classification techniques. Our main contribution is the introduction of a ”confidence” level for each rule. Initially, we define the ”representativeness” of a data point as the distance from its class’s center. Afterward, each rule’s confidence is calculated based on representativeness of data points it covers. This confidence is incorporated into the initialization of the corresponding Mass Assignment Function (MAF), providing a better starting point for the DSGD’s optimizer and enabling faster, more effective convergence. The code is available at https://github.com/HaykTarkhanyan/DSGD-Enhanced.

HTML

PDF

DOI: 10.3897/jucs.164654

Authors: Wolfram Luther, A. J. Han Vinck

Abstract: This volume presents a conference paper selection from the 4th Workshop on Collaborative Technologies and Data Science in Smart City Applications (CODASSCA 2024): Data Science and Reliable Machine Learning, held in Yerevan, Armenia, October 3-6, 2024, https://codassca2024.aua.am/. The special issues guest editors invited five groups of authors from Armenia, Chile, Germany, the UK, and the USA to submit enlarged versions of their CODASSCA 2024 papers There was also a J.UCS open call so that any author could submit papers on the highlighted subjects. The invitation to review the 16 contributions received was accepted by 16 experts, and, after three rounds, seven articles were finally accepted for publication in the special issue.

HTML

PDF

DOI: 10.3897/jucs.101550

Authors: Anita Herrera, Ángel Arroyo, Alfredo Jiménez, Álvaro Herrero

Abstract: The analysis of Artificial Intelligence techniques and models used in the tourism sector provides insightful information for the management and innovation of this industry. In this paper, we conduct a comprehensive review of the different techniques and models, in regards to Artificial Intelligence when applied to the tourism industry. Specifically, we present a categorization of Artificial Intelligence applications used in different areas of tourism. The results allow to recognize valid studies and useful tools for the activation and growth of the tourism sector, an industry that represents a significant increase in the Gross Domestic Product of various economies and supports the development of life conditions for their inhabitants. Artificial Intelligence applications generate more personalized travel experiences, improve the efficiency of tourism services and strengthen the tourism competitiveness of the destination.

HTML

XML

PDF

DOI: 10.3897/jucs.112797

Authors: Lorenz Gillner, Ekaterina Auer

Abstract: In this paper, we describe the requirements for traceable open-source data retrieval in the context of computation of BRCA1/2 mutation probabilities (mutations in two tumor-suppressor genes responsible for hereditary BReast or/and ovarian CAncer). We show how such data can be used to develop a Dempster-Shafer model for computing the probability of BRCA1/2 mutations enhanced by taking into account the actual age of a patient or a family member in an appropriate way even if it is not known exactly. The model is compared with PENN II and BOADICEA (based on undisclosed data), two established platforms for this purpose accessible online, as well as with our own previous models. A proof-of-concept implementation shows that set-based techniques are able to provide better information about mutation probabilities, simultaneously highlighting the necessity for ground truth data of high quality.

HTML

XML

PDF

DOI: 10.3897/jucs.69619

Authors: Vít Novotný, Michal Štefánik, Eniafe Festus Ayetiran, Petr Sojka, Radim Řehůřek

Abstract: In 2018, Mikolov et al. introduced the positional language model, which has characteristics of attention-based neural machine translation models and which achieved state-of-the-art performance on the intrinsic word analogy task. However, the positional model is not practically fast and it has never been evaluated on qualitative criteria or extrinsic tasks. We propose a constrained positional model, which adapts the sparse attention mechanism from neural machine translation to improve the speed of the positional model. We evaluate the positional and constrained positional models on three novel qualitative criteria and on language modeling. We show that the positional and constrained positional models contain interpretable information about the grammatical properties of words and outperform other shallow models on language modeling. We also show that our constrained model outperforms the positional model on language modeling and trains twice as fast.

HTML

XML

PDF

DOI: 10.3217/jucs-023-07-0673

Authors: Yanxia Sun, Zenghui Wang

Abstract: To improve the optimization performance of multi-objective particle swarm optimization, a new sub-swarm method, where the particles are divided into several sub-swarms, is proposed. To enhance the quality of the Pareto front set, a new adaptive sharing scheme, which depends on the distances from nearest neighbouring individuals, is proposed and applied. In this method, the first sub-swarms particles dynamically search their corresponding areas which are around some points of the Pareto front set in the objective space, and the chosen points of the Pareto front set are determined based on the adaptive sharing scheme. The second sub-swarm particles search the rest objective space, and they are away from the Pareto front set, which can promote the global search ability of the method. Moreover, the core points of the first sub-swarms are dynamically determined by this new adaptive sharing scheme. Some Simulations are used to test the proposed method, and the results show that the proposed method can achieve better optimization performance comparing with some existing methods.

HTML

XML

PDF

DOI: 10.3217/jucs-023-04-0353

Authors: Aleksei Deon, Yulian Menyaev

Abstract: Twisting generators for pseudorandom numbers may use a congruential array to simulate stochastic sequences. Typically, the computer program controls the quantity of elements in array to limit the random access memory. This technique may have limitations in situations where the stochastic sequences have an insufficient size for some application tasks, ranging from theoretical mathematics and technic constructions to biological and medical studies. This paper proposes a novel approach to generate complete stochastic sequences which don’t need a congruential twisting array. The results of simulation confirm that received random numbers are distributed absolutely uniformly in the set of unique sequences. Moreover, combination of this novel approach with an algorithm of tuning for twisting generation affords the length extension of created sequences without requiring additional computer random access memory.

HTML

XML

PDF

DOI: 10.3217/jucs-022-08-1023

Authors: Aleksei Deon, Yulian Menyaev

Abstract: Random sequences are widely used in theoretical and practical areas of interests in human and technical activities. An important part of these fields is referred to as the procedures of producing stochastic values. One direction adapts the sequenced generating of pseudorandom numbers and the other direction uses all stochastic sequences in objects of completed sets. The first direction is well studied and is traditionally used in cryptography and technical systems in medical and biological trials. The second direction is generally used in systems for preliminary universal testing where all or characteristically important sequences belong to a given diapason of actions are required. In this current work we explore the second direction, where the underlying approaches in modern generators of random numbers are considered. The simulation of complete sets of random numbers shows that either skipping or repeating of generated values is possible. We've formed the requirements that if followed, the problems of skipping and repeating are overcome. Next, we’ve proposed novel algorithms to form completed ranked sets of random sequences. Also, we've proposed novel algorithms on the basis of factorial expansion of random numbers which provide fast generation of such sequences. A discussion of the advantages and disadvantages of the indicated statements completes this paper.

HTML

XML

PDF

DOI: 10.3217/jucs-021-13-1790

Authors: Ana Freire, Craig Macdonald, Nicola Tonellotto, Iadh Ounis, Fidel Cacheda

Abstract: Large-scale search engines are built upon huge infrastructures involvingthousands of computers in order to achieve fast response times. In contrast, the energy consumed (and hence the financial cost) is also high, leading to environmental damage. This paper proposes new approaches to increase energy and financial savings in large-scale search engines, while maintaining good query response times. We aim to improve current state-of-the-art models used for balancing power and latency, by integratingnew advanced features. On one hand, we propose to improve the power savings by completely powering down the query servers that are not necessary when the load ofthe system is low. Besides, we consider energy rates into the model formulation. On the other hand, we focus on how to accurately estimate the latency of the whole systemby means of Queueing Theory. Experiments using actual query logs attest the high energy (and financial) savingsregarding current baselines. To the best of our knowledge, this is the first paper in successfully applying stationary Queueing Theory models to estimate the latency in alarge-scale search engine.

HTML

XML

PDF

DOI: 10.3217/jucs-021-06-0842

Authors: Héctor Allende-Cid, Héctor Allende, Raúl Monge, Claudio Moraga

Abstract: When distributed data sources have different contexts the problem of Distributed Re-gression becomes severe. It is the underlying law of probability that constitutes the context of a source. A new Distributed Regression System is presented, which makes use of a discrete rep-resentation of the probability density functions (pdfs). Neighborhoods of similar datasets are detected by comparing their approximated pdfs. This information supports an ensemble-basedapproach, and the improvement of a second level unit, as it is the case in stacked generalization. Two synthetic and six real data sets are used to compare the proposed method with otherstate-of-the-art models. The obtained results are positive for most datasets.

HTML

XML

PDF

DOI: 10.3217/jucs-020-15-2032

Authors: Antonio Cabot, Eva Garcia-Lopez, Luis De-Marcos, Javier Abraham-Curto

Abstract: E-learning has been a revolution in recent years in training field. This, combined with the increased use of mobile devices has caused the emergence of m-learning. Hence new problems have appeared in the training field, such as displaying correctly learning contents in a mobile device that has restricted features or taking into account the learner's context in the learning process, who could be anywhere. For this reason the adaptation concept is used, in order to personalize or adapt the learning contents to each student. This paper presents a case study in a real course using a multi-agent system for adapting the learning contents to the learner's context and to his/her mobile device. The results of the experiment show that the students who used the adaptive system (experimental group) obtained better grades than the students who did not (control group).

HTML

XML

PDF

DOI: 10.3217/jucs-020-05-0615

Authors: Paola Accioly, Paulo Borba, Rodrigo Bonifacio

Abstract: SPL testing has been considered a challenging task, mainly due to the diversity of products that might be generated from an SPL. To deal with this problem, techniques for specifying and deriving product specific functional test cases have been proposed. However, there is little empirical evidence of the benefits and drawbacks of these techniques. To provide this kind of evidence, in a previous work we conducted an empirical study that compared two design techniques for black-box manual testing, a generic technique that we have observed in an industrial test execution environment, and a product specific technique whose functional test cases could be derived using any SPL approach that considers variations in functional tests. Besides revisiting the first study, here we present a second study that reinforce our findings and brings new insights to our investigation. Both studies indicate that specific test cases improve test execution productivity and quality.

HTML

XML

PDF

DOI: 10.3217/jucs-016-01-0121

Authors: Özgür Kabak, Füsun Ülengin

Abstract: Several supply chain and production planning models in the literature assume the demands are fuzzy but most of them do not offer a specific technique to derive the fuzzy demands. In this study, we propose a methodology to obtain a fuzzy-demand forecast that is represented by a possibilistic distribution. The fuzzy-demand forecast is found by aggregating forecasts based on different sources; namely statistical forecasting methods and experts’ judgments. In the methodology, initially, the forecast derived from the statistical forecasting techniques and experts’ judgments are represented by triangular possibilistic distributions. Subsequently, those results are combined by using weights assigned to each of them. A new objective weighting approach is used to find the weights. The proposed methodology is illustrated by an example and a sensitivity analysis is provided.

HTML

XML

PDF

DOI: 10.3217/jucs-015-13-2608

Authors: Vitoantonio Bevilacqua, Lucia Cariello, Marco Giannini, Giuseppe Mastronardi, Vito Santarcangelo, Rocco Scaramuzzi, Antonella Troccoli

Abstract: This paper describes a comparative study between an Artificial Neural Network (ANN) and a geometric technique to detect for biometric applications,the bifurcation points of blood vessels in the retinal fundus. The first step is an image preprocessing phase to extract retina blood vessels. The contrast of the blood vessels from the retinal image background is enhanced in order to extract the blood vessels skeleton. Successively, candidate points of bifurcation are individualized by approximating the skeleton lines in segments. The distinction between bifurcations and vessel bends is carried out through the employment of two methods: geometric (through the study of intersections within the region obtained thresholding the image portion inside a circle centered around the junctions point and the circumference of the same circle) and an ANN. The results obtained are compared and discussed.

HTML

XML

PDF

DOI: 10.3217/jucs-015-04-0826

Authors: Vitoantonio Bevilacqua, Giuseppe Mastronardi, Filippo Menolascina, Paolo Pannarale, Giuseppe Romanazzi

Abstract: Inferring gene regulatory networks from data requires the development of algorithms devoted to structure extraction. When time-course data is available, gene interactions may be modeled by a Bayesian Network (BN). Given a structure, that models the conditional independence between genes, we can tune the parameters in a way that maximize the likelihood of the observed data. The structure that best fit the observed data reflects the real gene network's connections. Well known learning algorithms (greedy search and simulated annealing) devoted to BN structure learning have been used in literature. We enhanced the fundamental step of structure learning by means of a classical evolutionary algorithm, named GA (Genetic algorithm), to evolve a set of candidate BN structures and found the model that best fits data, without prior knowledge of such structure. In the context of genetic algorithms, we proposed various initialization and evolutionary strategies suitable for the task. We tested our choices using simulated data drawn from a gene simulator, which has been used in the literature for benchmarking [Yu et al. (2002)]. We assessed the inferred models against this reference, calculating the performance indicators used for network reconstruction. The performances of the different evolutionary algorithms have been compared against the traditional search algorithms used so far (greedy search and simulated annealing). Finally we individuated as best candidate an evolutionary approach enhanced by Crossover-Two Point and Selection Roulette Wheel for the learning of gene regulatory networks with BN. We show that this approach outperforms classical structure learning methods in elucidating the original model of the simulated dataset. Finally we tested the GA approach on a real dataset where it reach 62% of recovered connections (sensitivity) and 64% of direct connections (precision), outperforming the other algorithms.

HTML

XML

PDF

DOI: 10.3217/jucs-015-02-0488

Authors: Chin-Ling Chen

Abstract: This study has proposed a new detection method for DDoS attack traffic based on two-sample t-test. We first investigate the statistics of normal SYN arrival rate (SAR) and confirm it follows normal distribution. The proposed method identifies the attack by testing 1) the difference between incoming SAR and normal SAR, and 2) the difference between the number of SYN and ACK packets. The experiment results show that the possibilities of both false positives and false negatives are very low. The proposed mechanism is also demonstrated to have the capability of detecting DDoS attack quickly.

HTML

XML

PDF

DOI: 10.3217/jucs-013-11-1655

Authors: Gabriel Istrate

Abstract: The distribution of overlaps of solutions of a random constraint satisfaction problem (CSP) is an indicator of the overall geometry of its solution space. For random k-SAT, nonrigorous methods from Statistical Physics support the validity of the one step replica symmetry breaking approach. Some of these predictions were rigorously confirmed in [Mézard et al. 2005a] [Mézard et al. 2005b]. There it is proved that the overlap distribution of random k-SAT, k ≥ 9, has discontinuous support. Furthermore, Achlioptas and Ricci-Tersenghi [Achlioptas and Ricci-Tersenghi 2006] proved that, for random k-SAT, k ≥ 8, and constraint densities close enough to the phase transition: - there exists an exponential number of clusters of satisfying assignments. - the distance between satisfying assignments in different clusters is linear. We aim to understand the structural properties of random CSP that lead to solution clustering. To this end, we prove two results on the cluster structure of solutions for binary CSP under the random model from [Molloy 2002]: 1. For all constraint sets S (described in [Creignou and Daudé 2004, Istrate 2005]) such that SAT (S) has a sharp threshold and all q ∈ (0, 1], q-overlap-SAT (S) has a sharp threshold. In other words the first step of the approach in [Mézard et al. 2005a] works in all nontrivial cases. 2. For any constraint density value c < 1, the set of solutions of a random instance of 2-SAT form with high probability a single cluster. Also, for and any q ∈ (0, 1] such an instance has with high probability two satisfying assignment of overlap ~ q. Thus, as expected from Statistical Physics predictions, the second step of the approach in [Mézard et al. 2005a] fails for 2-SAT.

HTML

XML

PDF

DOI: 10.3217/jucs-012-09-1332

Authors: Ferenc Nagy

Abstract: As we know the Cauchy distribution plays an important role in Probability Theory and Statistics. In this paper, we investigate the estimation of the location and the scale parameter. Both the one-dimensional problem and the multidimensional problem are studied for large sample. In the one-dimensional case, we give two algorithms for the estimation. The first one is an iterative method for which we prove the convergence and we show that the rate of convergence is geometric. The second algorithm provides an exact solution to the problem. In the multidimensional case, we give an algorithm analogous to the one-dimensional case. Computer experiments show that the rate of convergence is similar to the one-dimensional iterative algorithm.

HTML

XML

PDF

DOI: 10.3217/jucs-012-06-0710

Authors: Beate Bollig

Abstract: Combinatorial property testing, initiated formally by Goldreich, Goldwasser, and Ron in [Goldreich et al. (1998)] and inspired by Rubinfeld and Sudan in [Rubinfeld and Sudan 1996], deals with the relaxation of decision problems. Given a property P the aim is to decide whether a given input satisfies the property P or is far from having the property. A property P can be described as a language, i.e., a nonempty family of binary words. The associated property to a family of boolean functions f = (fn) is the set of 1-inputs of f. By an attempt to correlate the notion of testing to other notions of low complexity property testing has been considered in the context of formal languages. Here, a brief summary of results on testing properties defined by formal languages and by languages implicitly represented by small restricted branching programs is provided.

HTML

XML

PDF

DOI: 10.3217/jucs-012-06-0672

Authors: Makoto Matsumoto, Mutsuo Saito, Hiroshi Haramoto, Takuji Nishimura

Abstract: Pseudorandom number generators are widely used in the area of simulation. Defective generators are still widely used in standard library programs, although better pseudorandom number generators such as the Mersenne Twister are freely available. This manuscript gives a brief explanation on pseudorandom number generators for Monte Carlo simulation. The existing definitions of pseudorandomness are not satisfactorially practical, since the generation of sequences satisfying the definitions is sometimes impossible, somtimes rather slow. As a compromise, to design a fast and reliable generator, some mathematical indices are used as measures of pseudorandomness, such as the period and the higher-dimensional equidistribution property. There is no rigorous justification for the use of these indices as measures of pseudorandomness, but experiences show their usefulness in choosing pseudorandom number generators.

HTML

XML

PDF

DOI: 10.3217/jucs-011-10-1718

Authors: Christel Baier

Abstract: Constraint automata have been used as an operational model for Reo which offers a channel-based framework to compose complex component connectors. In this paper, we introduce a variant of constraint automata with discrete probabilities and nondeterminism, called probabilistic constraint automata. These can serve for compositional reasoning about connector components, modelled by Reo circuits with unreliable channels, e.g., that might lose or corrupt messages, or channels with random output values that, e.g., can be helpful to model randomized coordination principles.

HTML

XML

PDF

DOI: 10.3217/jucs-009-10-1196

Authors: Cristian S. Calude, Elena Calude, Michael Dinneen

Abstract: For almost 350 years it was known that 1729 is the smallest integer which can be expressed as the sum of two positive cubes in two different ways. Motivated by a famous story involving Hardy and Ramanujan, a class of numbers called Taxicab Numbers has been defined: Taxicab(k, j, n) is the smallest number which can be expressed as the sum of j kth powers in n different ways. So, Taxicab(3, 2, 2) = 1729, Taxicab(4, 2, 2) = 635318657. Computing Taxicab Numbers is challenging and interesting, both from mathematical and programming points of view. The exact value of Taxicab(6) = Taxicab(3, 2, 6) is not known, however, recent results announced by Rathbun [R2002] show that Taxicab(6) is in the interval [10 18 , 24153319581254312065344]. In this note we show that with probability greater than 99%, Taxicab(6) = 24153319581254312065344.

HTML

XML

PDF

DOI: 10.3217/jucs-006-01-0212

Authors: Kundi Xue, Marius Zimand

Abstract: Extractors are a special type of binary graphs that can be utilized to improve the quality of randomness sources that generate strings with small entropy. The paper explores constructions of extractors that are practical and easy to implement. Randomized and deterministic constructions are presented and compared with some previously known constructions that achieve very good asymptotical performances. One of our methods is shown to have a better behavior for reasonable values of the involved parameters. 1 C.S.Calude and G.Stefanescu (eds.). Automata, Logic, and Computability. Special issue dedicated to Professor Sergiu Rudeanu Festschrift.

HTML

XML

PDF

DOI: 10.3217/jucs-005-10-0633

Authors: Winfrid Schneeweiss

Abstract: Fault trees show which joint components' faults mean system faults. Fault trees can often be used to determine dependability parameters of systems. Here it is shown that i) binary decision diagrams (BDDs) can also be used to calculate system mean failure frequency, ii) modeling dynamics of fault trees does not always mean Markov modeling, iii) a deeper understanding of interrelations between s-dependent components is supported, rather, by Petri nets than by state transition graphs.

HTML

XML

PDF

DOI: 10.3217/jucs-005-08-0471

Authors: Ileana Popescu, Monica Dumitrescu

Abstract: Laha distribution has been introduced in 1958 as an example of a non-normal distribution where the quotient follows the Cauchy law. In this paper we present two procedures for the computer generation of this distribution and we discuss its applications to life time modelling.

HTML

XML

PDF

DOI: 10.3217/jucs-004-08-0675

Authors: Ville Leppänen

Abstract: We present a novel approach to parallel computing, where (virtual) PRAM processors are represented as light-weight threads, and each physical processor is capable of managing several threads. Instead of moving read and write requests, and replies between processor&memory pairs (and caches), we move the light-weight threads. Consequently, the processor load balancing problem reduces to the problem of producing evenly distributed memory references. In PRAM computations, this can be achieved by properly hashing the shared memory into the processor&memory pairs. We describe the idea of moving threads, and show that the moving threads framework provides a natural validation for Brent's theorem in work-optimal PRAM simulation situations on mesh of trees, coated mesh, and OCPC based distributed memory machines (DMMs). We prove that an EREW PRAM computation requiring work W and time T, can be implemented work-optimally on those p-processor DMMs with high probability, if W = , where D is the diameter of the underlying routing machinery. The computation is work-optimal regardless how (virtual) PRAM processors terminate or initiate new PRAM processors during the computation. Our result is based on using only one randomly chosen hash function and on showing, how the threads (PRAM processors) can spawn new threads in required time on p-processor OCPC, 2-dimensional mesh of trees, 2-dimensional coated, and 3-dimensional coated mesh. A deterministic spawning algorithm is provided for all cases, although a randomized algorithm would be sufficient due to the randomized nature of time-processor optimal PRAM simulations.

HTML

XML

PDF

DOI: 10.3217/jucs-003-07-0821

Authors: Peter Undrill, Kostas Delibasis, George Cameron

Abstract: Stack filters are a class of non-linear spatial operators used for suppression of noise in signals. In this work their design is formulated as an optimisation problem and a method that uses Genetic Algorithms (GAs) to perform the configuration is explained. Because of its computational complexity the process has been implemented as a distributed parallel GA using the Parallel Virtual Machine (PVM) software. We present the results of applying our stack filters to the restoration of magnetic resonance (MR) images corrupted with uniform, uncorellated, noise showing improved statistical performance compared with the median filter and indicating better retention of image details. The efficiency of the parallel implementation is examined, addressing both algorithmic and data decomposition, showing that execution times can be significantly reduced by distributing the task across a network of heterogeneous processors.

HTML

XML

PDF

DOI: 10.3217/jucs-001-01-0048

Authors: Cristian S. Calude

Abstract: Chaitin s algorithmic definition of random strings - based on the complexity induced by self-delimiting computers - is critically discussed. One shows that Chaitin s model satisfy many natural requirements related to randomness, so it can be considered as an adequate model for finite random objects. It is a better model than the original (Kolmogorov) proposal. Finally, some open problems will be discussed.

HTML

XML

PDF

DOI: 10.3217/jucs-001-01-0023

Authors: Ville Leppänen

Abstract: We show how to implement an -processor EREW PRAM workoptimally on a 2-dimensional n-sided mesh of trees, consisting of n processors, n memory modules, and nodes. Similarly, we prove that an -processor EREW PRAM can be implemented work-optimally on a 3-dimensional n-sided mesh of trees. By the work-optimality of implementations we mean that the expected routing time of PRAM memory requests is per simulated PRAM processor with high probability. Experiments show that on relatively small and the cost per simulated PRAM processor is 1.5-2.5 in the 2-dimensional case, and 2-3 in the 3-dimensional case. If at each step at most 1/3'th of the PRAM processors make a reference to the shared memory, then the simulation cost is approximately 1. We also compare our work-optimal simulations to those proposed for coated meshes.

HTML

XML

PDF