Papers by Fabiana Piccoli
Este trabajo aborda las líneas de trabajo a seguir, con el objetivo de desarrollar e incluir técn... more Este trabajo aborda las líneas de trabajo a seguir, con el objetivo de desarrollar e incluir técnicas de Computación de Alto Desempeño en problemas con datos masivos o problemas Big-Data. El desarrollo de la infraestructura, metodologías y herramientas de Computación de Alto Desempeño y su empleo eficiente en la solución de problemas que tratan grandes cantidades de datos, debe considerar la naturaleza de estos.
Maria A. Murazzo, Maria Fabiana Piccoli, Nelson R. Rodriguez, Diego Medel, Jorge N. Mercado, Fede... more Maria A. Murazzo, Maria Fabiana Piccoli, Nelson R. Rodriguez, Diego Medel, Jorge N. Mercado, Federico Sanchez, Ana Laura Molina, Martin Tello Departamento de Informática – FCEFy N, UNSJ. Departamento de Informática – FCFMy N, UNSL. Departamento de Matemática – FI, UNSJ. Alumno Avanzado de la Carrera Licenciatura en Ciencias de la Computación. Alumno Avanzado de la Carrera Licenciatura en Sistemas de Información.
XIX Workshop de Investigadores en Ciencias de la Computación (WICC 2017, ITBA, Buenos Aires), Apr 1, 2017
Informatore agrario, 2002
Acta Botanica Neerlandica, 1995
ABSTRACT SUMMARY The development fruits (nutlets) and seeds of Schoenus nigricans, a perennial se... more ABSTRACT SUMMARY The development fruits (nutlets) and seeds of Schoenus nigricans, a perennial sedge of wet coastal dune slacks, was compared between populations of the Mediterranean Sea and the North Sea, to examine the hypothesis of adaptation in a geographical gradient (Baker 1972). The development period of nutlets and seeds lasted several months. The serial insertion of nutlets on the spikelet caused a negative mass gradient from adaxial to abaxial nutlets. Final nutlet mass (first adaxial spikelet position) of Mediterranean plants was 30% higher than that of North Sea plants, extending Baker's hypothesis of the inverse relationship between seed (nutlet) mass and geographical altitude to geographical latitude. The difference in final nutlet mass between plants from populations of the Mediterranean and North Sea persisted, when plants were grown under the climatic conditions of the North Sea plants. However, the nutlet mass, but not the seed mass of the Mediterranean plants, grown under Dutch climatic conditions, was less than at their original size. Start of flowering of the Mediterranean plants under Dutch climatic conditions was delayed by 1–2 months, compared with their flowering at the Mediterranean sites. Nutlets of Mediterranean plants ripened later than those of Dutch plants under the Dutch climatic conditions. The differentiation in nutlet and seed masses and plant height in the geographic gradient in Europe and its persistence under experimental conditions indicates a strong genetic component of population differentiation. The data do not support the hypothesis of ‘general-purpose genotypes’ (Schmid 1992) in this wet dune slack species.
High Performance Computing Symposium, 2011
IASTED PDCS, 2005
ABSTRACT Though there have been substantial progress, the current status of parallel computing is... more ABSTRACT Though there have been substantial progress, the current status of parallel computing is still inmature. No single model of parallel computation has yet come to dominate developments in parallel computing in the way that the von Neumann model ...
Portal de Libros de la Universidad Nacional de La Plata, 2019
Quantum Penny Flip: an open source serious game for quantum computing assessment Zapirain, Esteba... more Quantum Penny Flip: an open source serious game for quantum computing assessment Zapirain, Esteban A. (1); Massa, Stella M. (2); Cardoso, Juan P. (3) (1) (2) (3) Grupo de Investigación en Tecnologías Interactivas. (1) (2) (3)
IEEE International Conference on Cloud Computing Technology and Science, 2018
Los avances tecnológicos han permitido que se generen grandes cantidades de datos, los cuales nec... more Los avances tecnológicos han permitido que se generen grandes cantidades de datos, los cuales necesitan ser almacenados y procesados de manera eficiente. Surge así el paradigma Big Data, donde el principal requerimiento no solo es la capacidad de cómputo, sino el manejo en un tiempo razonable de ingentes cantidades de datos. En este contexto, las aplicaciones para big data necesitan ser escalables, livianas, autocontenidas, distribuidas y replicadas con el objetivo de lograr la mejor performance frente a variaciones del volumen de datos. Para lograr esto, este trabajo propone ajustar la construcción de aplicaciones a una arquitectura basada en microservicios los cuales puedan ser implementados con contenedores. La replicación y distribución para lograr altos niveles de escalabilidad se plantea mediante la orquestación de contenedores sobre una arquitectura distribuida virtualizada.
Procedia Computer Science, 2013
Visualization methods of medical imagery based on volumetric data constitute a fundamental tool f... more Visualization methods of medical imagery based on volumetric data constitute a fundamental tool for medical diagnosis, training and pre-surgical planning. Often, large volume sizes and/or the complexity of the required computations present serious obstacles for reaching higher levels of realism and real-time performance. Performance and efficiency are two critical aspects in traditional algorithms based on complex lighting models. To overcome these problems, a volume rendering algorithm, PD-Render intra for individual networked nodes in a parallel distributed architecture with a single GPU per node is presented in this paper. The implemented algorithm is able to achieve photorealistic rendering as well as a high signal-tonoise ratio at interactive frame rates. Experiments show excellent results in terms of efficiency and performance for rendering medical volumes in real time.
XIII Congreso Argentino de Ciencias de la Computación, 2007
Images can reveal useful information to human users when are analyzed. The explosive growth in ap... more Images can reveal useful information to human users when are analyzed. The explosive growth in applying images as data in many fields of science, business, medicine, etc, demands greater processing power. With the advances in multimedia data acquisition and storage techniques, the need for automatically discovering knowledge from large image collections is becoming more and more relevant. Image mining, a relatively new and very promising field of investigation, tries to ease this problem proposing some solutions for the extraction of significant and potentially useful patterns from these tremendous data volume. This research field implies different stages, most of them demanding so many resources and computational time. The use of parallel computation is a good starting-point. Image mining process appears to be algorithmically complex requiring computing power levels that only parallel paradigms can provide in a timely way. As data sets involved are large, rapidly growing larger and images provide a natural source of parallelism, parallels computers could be organized to handle such big collection effectively. At this work we will examine the image mining problem with its computational cost, propose a possible global or local parallel solution and also identify some future research directions for image mining parallelism.
Mecánica Computacional, 2008
An Image Mining System (IMS) requires real time processing often using special purpose hardware. ... more An Image Mining System (IMS) requires real time processing often using special purpose hardware. The work herein presented refers to the application of cluster computing for on line image processing inside an IMS, where the end user benefits from the operation on data with a high degree locality and parallelism. The virtual parallel computer is composed by a cluster of personal computers connected by a low cost network. The aim is to minimise the processing time of a high level image processing package. The image processing function set developed to manage the parallel execution is described and some results obtained from the parallelisation of image processing algorithms are discussed.
Communications in computer and information science, 2022
Journal of computer science and technology, Dec 12, 2018
As computer networks have transformed in essential tools, their security has become a crucial pro... more As computer networks have transformed in essential tools, their security has become a crucial problem for computer systems. Detecting unusual values from large volumes of information produced by network traffic has acquired huge interest in the network security area. Anomaly detection is a starting point to prevent attacks, therefore it is important for all computer systems in a network have a system of detecting anomalous events in a time near their occurrence. Detecting these events can lead network administrators to identify system failures, take preventive actions and avoid a massive damage. This work presents, first, how identify network traffic anomalies through applying parallel computing techniques and Graphical Processing Units in two algorithms, one of them a supervised classification algorithm and the other based in network traffic image processing. Finally, it is proposed as a challenge to resolve the anomalies detection using an unsupervised algorithm as Deep Learning.
XIV Congreso Argentino de Ciencias de la Computación, 2008
Increasing amount of image data transmitted via Internet has triggered the development of general... more Increasing amount of image data transmitted via Internet has triggered the development of general purposes Image Mining Systems (IMS). An IMS performance relies on a good and fast feature vector specification that describes univocally an entire image. Vector size and the relationship between each evaluated feature and its computation time are critical, moreover when the image amount is big enough. Decreasing this IMS computational complexity by means of parallelism at the different involved tasks is one solution. Nowadays clusters of computers are already widely used as a low cost and high utility option to special-purpose machines, and suited to solve image processing problems with a high degree of data locality and parallelism. At this paper, we will focus on parallelism into the IMS processing stage trying to accelerate the feature vector calculus thru a cluster architecture attempting to give a better performance to the whole image mining system.
Communications in computer and information science, 2018
Detecting unusual values from large volumes of information produced by network traffic has acquir... more Detecting unusual values from large volumes of information produced by network traffic has acquired considerable interest in the network security area. Having a system of detecting anomalous events in a time near their occurrence, it is important for all computer systems in a network. Detecting anomalous values can lead network administrators to identify system failures, take preventative actions and avoid a massive spread. Anomaly detection is a starting point to prevent attacks. In this article, we present a form of data preprocessing to identify anomalies using a supervised classification algorithm, image processing, parallel computing techniques and Graphical Processing Units.
XXIII Congreso Argentino de Ciencias de la Computación (La Plata, 2017)., Oct 1, 2017
D e te c ta r v a lo re s a n o rm a le s a p a r tir d e g ra n d e s v o lú m e n e s d e in fo... more D e te c ta r v a lo re s a n o rm a le s a p a r tir d e g ra n d e s v o lú m e n e s d e in fo rm a c ió n p r o d u c id o p o r e l trá fic o d e r e d h a a d q u irid o u n in te ré s c o n s id e ra b le e n e l á re a d e s e g u rid a d d e re d e s. E s d e re le v a n c ia p a r a to d o s is te m a d e c o m p u ta d o ra s c o n e c ta d a s a u n a r e d c o n ta r c o n u n s is te m a d e d e te c c ió n d e e v e n to s a n ó m a lo s y u n tie m p o d e o b te n c ió n d e ta le s e v e n to s lo m á s c e rc a n o p o s ib le a s u o c u rre n c ia. D e te c ta r v a lo re s a n ó m a lo s p u e d e c o n d u c ir a lo s a d m in is tra d o re s d e r e d a id e n tific a r f a lla s d e l s is te m a y , p o r lo ta n to , to m a r m e d id a s p r e v e n tiv a s a n te s d e u n a m a s iv a p ro p a g a c ió n. L a d e te c c ió n d e a n o m a lía s e s u n p u n to d e p a rtid a p a r a e v ita r n u e v o s a ta q u e s. E n e ste a rtíc u lo , p re s e n ta m o s u n a fo rm a d e p rep ro c e s a r d a to s p a r a id e n tific a r a n o m a lía s m e d ia n te u n a lg o ritm o d e c la s ific a c ió n K-N N c o n té c n ic a s d e c o m p u ta c ió n p a ra le la s u s a n d o U n id a d e s d e P ro c e s a m ie n to G rá fic o .
Uploads
Papers by Fabiana Piccoli