Search

Developing and optimizing software applications for high performance and energy efficiency is a very challenging task, even when considering a single target machine. For instance, optimizing for multicore-based computing systems requires in-depth knowledge about programming languages, application programming interfaces, compilers, performance tuning tools, and computer architecture and organization. Many of the tasks of performance engineering methodologies require manual efforts and the use of different tools not always part of an integrated toolchain. This paper presents Pegasus, a performance engineering approach supported by a framework that consists of a source-to-source compiler, controlled and guided by strategies programmed in a Domain-Specific Language, and an autotuner. Pegasus is a holistic and versatile approach spanning various decision layers composing the software stack, and exploiting the system capabilities and workloads effectively through the use of runtime autotuning. The Pegasus approach helps developers by automating tasks regarding the efficient implementation of software applications in multicore computing systems. These tasks focus on application analysis, profiling, code transformations, and the integration of runtime autotuning. Pegasus allows developers to program their strategies or to automatically apply existing strategies to software applications in order to ensure the compliance of non-functional requirements, such as performance and energy efficiency. We show how to apply Pegasus and demonstrate its applicability and effectiveness in a complex case study, which includes tasks from a smart navigation system.

The LEXIS (Large-scale EXecution for Industry & Society) Weather and Climate Pilot is developing a system for the prediction of water-related phenomena and their associated socioeconomic impacts. The system is based on multiple models chained together, as global weather models, high-resolution regional weather models, domain-specific application models (hydrological and forest fire risk forecasts), and impact models providing information (such as air quality and rainfall intensity) for key decisions and policymakers. This chapter describes the key aspect of this pilot in terms of serving model output data and products with cloud and high-performance data analytics environments, on top of a Weather Climate Data API, as well as the porting of models on the LEXIS infrastructure via different virtualization strategies (as virtual machine and dockers).

The LEXIS project (Large-scale EXecution for Industry & Society, H2020 GA825532) provides a platform for optimised execution of Cloud-HPC workflows, reducing computation time and increasing energy efficiency. The system will rely on advanced, distributed orchestration solutions (Atos YSTIA Suite, with Alien4Cloud and Yorc, based on TOSCA), the High-End Application Execution Middleware HEAppE, and new hardware capabilities for maximising efficiency in data processing, analysis and transfer (e.g. Burst Buffers with GPU- and FPGA-based data reprocessing). LEXIS handles computation tasks and data from three Pilots, based on representative and demanding HPC/Cloud-Computing use cases in Industry (SMEs) and Science: i) Simulations of complex turbomachinery and gearbox systems in Aeronautics, ii) Tsunami simulations and earthquake loss assessments which are time-constrained to enable immediate warnings and to support well-informed decisions, and iii) Weather and Climate simulations where massive amounts of in-situ data are assimilated to improve forecasts. A user-friendly LEXIS web portal, as a unique entry point, will provide access to data as well as workflow-handling and remote visualisation functionality. As part of its back-end, LEXIS builds an elaborate system for the handling of input, intermediate and result data. At its core, a Distributed Data Infrastructure (DDI) ensures the availability of LEXIS data at all participating HPC sites, which will be federated with a common LEXIS Authentication and Authorisation Infrastructure (with unified security model, user database and policies). The DDI leverages best of breed data-management solutions from EUDAT, such as B2SAFE (based on iRODS) and B2HANDLE. REST APIs on top of it will ensure a smooth interaction with LEXIS workflows and the orchestration layer. Last, but not least, the DDI will provide functionalities for Research Data Management following the FAIR principles ("Findable, Interoperable, Accessible, Reusable"), e.g. DOI acquisition, which helps to publish and disseminate open data products.

Remote-sensing-driven urban change detection has been studied in many ways for decades for a wide field of applications, such as understanding socio-economic impacts, identifying new settlements, or analyzing trends of urban sprawl. Such kinds of analyses are usually carried out manually by selecting high-quality samples that binds them to small-scale scenarios, either temporarily limited or with low spatial or temporal resolution. We propose a fully automated method that uses a large amount of available remote sensing observations for a selected period without the need to manually select samples. This enables continuous urban monitoring in a fully automated process. Furthermore, we combine multispectral optical and synthetic aperture radar (SAR) data from two eras as two mission pairs with synthetic labeling to train a neural network for detecting urban changes and activities. As pairs, we consider European Remote Sensing (ERS-1/2) and Landsat 5 Thematic Mapper (TM) for 1991–2011 and Sentinel 1 and 2 for 2017–2021. For every era, we use three different urban sites—Limassol, Rotterdam, and Liège—with at least 500km2 each, and deep observation time series with hundreds and up to over a thousand of samples. These sites were selected to represent different challenges in training a common neural network due to atmospheric effects, different geographies, and observation coverage. We train one model for each of the two eras using synthetic but noisy labels, which are created automatically by combining state-of-the-art methods, without the availability of existing ground truth data. To combine the benefit of both remote sensing types, the network models are ensembles of optical- and SAR-specialized sub-networks. We study the sensitivity of urban and impervious changes and the contribution of optical and SAR data to the overall solution. Our implementation and trained models are available publicly to enable others to utilize fully automated continuous urban monitoring.

Traditional usage models of Supercomputing centres have been extended by High-Throughput Computing (HTC), High-Performance Data Analytics (HPDA) and Cloud Computing. The complexity of current compute platforms calls for solutions to simplify usage and conveniently orchestrate computing tasks. These enable also non-expert users to efficiently execute Big Data workflows. In this context, the LEXIS project (‘Large-scale EXecution for Industry and Society’, H2020 GA 825532, https://lexis-project.eu) sets up an orchestration platform for compute- and data-intensive workflows. Its main objective is to implement a front-end and interfaces/APIs for distributed data management and workflow orchestration. The platform uses an open-source Identity and Access Management solution and a custom billing system. The data management API allows data ingestion and staging between various infrastructures. The orchestration API allows execution of workflows specified in extended TOSCA. LEXIS uses innovative technologies like YORC and Alien4Cloud for orchestration or iRODS/EUDAT-B2SAFE for data management, accelerated by Burst Buffers. Three pilot use cases from Aeronautics Engineering, Earthquake/Tsunami Analysis, and Weather and Climate Prediction are used to test the services. On the road towards longer-term sustainability, we are expanding this user base and aiming at the immersion of more Supercomputing centres within the platform.

The present information society creates huge quantities of textual information. This information explosion is being handled using Information Retrieval Systems. Their tasks are effective storage and searching in the text collections. The amount of text stored in IRS and auxiliary data structures constitute a suitable material for data compression. However, the data that form the textual database of every IRS are very mixed and it is therefore useful to study special data compression methods. This chapter focuses on high compression ratio algorithms specialized in text compression in IRS that enable a fast decompression of individual documents, fully integrated with the IRS, and work with an adequate compression speed. These methods uses word-based compression methods combined with topical development of input documents. Experimental results prove that clustering has a positive impact on the compression ratio. The advantage of implementing this approach is that it is not necessary to change the existing compression algorithm. The only thing that changes is the ordering in which compressed documents are input. Decompression algorithms are not influenced at all and knowledge of topical development is not necessary.

We introduce KaraMIR, a musical project dedicated to karaoke song analysis. Within KaraMIR, we define Kara1k, a dataset composed of 1000 cover songs provided by Recisio Karafun application, and the corresponding 1000 songs by the original artists. Kara1k is mainly dedicated toward cover song identification and singing voice analysis. For both tasks, Kara1k offers novel approaches, as each cover song is a studio-recorded song with the same arrangement as the original recording, but with different singers and musicians. Essentia, harmony-analyser, Marsyas, Vamp plugins and YAAFE have been used to extract audio features for each track in Kara1k. We provide metadata such as the title, genre, original artist, year, International Standard Recording Code and the ground truths for the singer’s gender, backing vocals, duets, and lyrics’ language. KaraMIR project focuses on defining new problems and describing features and tools to solve them. We thus provide a comparison of traditional and new features for a cover song identification task using statistical methods, as well as the dynamic time warping method on chroma, MFCC, chords, keys, and chord distance features. A supporting experiment on the singer gender classification task is also proposed. The KaraMIR project website facilitates the continuous research.

High-Performance Big Data Analytics (HPDA) applications are characterized by huge volumes of distributed and heterogeneous data that require efficient computation for knowledge extraction and decision making. Designers are moving towards a tight integration of computing systems combining HPC, Cloud, and IoT solutions with artificial intelligence (AI). Matching the application and data requirements with the characteristics of the underlying hardware is a key element to improve the predictions thanks to high performance and better use of resources. We present EVEREST, a novel H2020 project started on October 1st, 2020 that aims at developing a holistic environment for the co-design of HPDA applications on heterogeneous, distributed, and secure platforms. EVEREST focuses on programmability issues through a data-driven design approach, the use of hardware-accelerated AI, and an efficient runtime monitoring with virtualization support. In the different stages, EVEREST combines state-of-the-art programming models, emerging communication standards, and novel domain-specific extensions. We describe the EVEREST approach and the use cases that drive our research., Comment: Paper accepted for presentation at the IEEE/EDAC/ACM Design, Automation and Test in Europe Conference and Exhibition (DATE 2021)

This paper is concerned with finding near optimal parameters for the inventory optimization model on large dataset. It is shown that our proposed method allows for very good model parameter estimation with great reduction in computation time. Model developed in cooperation with K2 atmitec s.r.o. company has four input parameters which must be set before the run. These parameters are estimated through computationally complex simulations by hyperparameter search. Since it is impossible to make grid search of the optimal parameters for all the input time series, it is necessary to approximate the parameters settings. This approximation is done through similarity search and computation of optimal parameters on the most central objects. Additionally, parameter estimation is improved by the clustering of time series and the results are upgraded by the new estimations.

The Sentinel-1 satellite system continuously observes European countries at a relatively high revisit frequency of six days per orbital track. Given the Sentinel-1 configuration, most areas in Czechia are observed every 1–2 days by different tracks in a moderate resolution. This is attractive for various types of analyses by various research groups. The starting point for interferometric (InSAR) processing is an original data provided in a Single Look Complex (SLC) level. This work represents advantages of storing data augmented to a specifically corrected level of data, SLC-C. The presented database contains Czech nationwide Sentinel-1 data stored in burst units that have been pre-processed to the state of a consistent well-coregistered dataset of SLC-C. These are resampled SLC data with their phase values reduced by a topographic phase signature, ready for fast interferometric analyses (an interferogram is generated by a complex conjugate between two stored SLC-C files). The data can be used directly into multitemporal interferometry techniques, e.g., Persistent Scatterers (PS) or Small Baseline (SB) techniques applied here. A further development of the nationwide system utilising SLC-C data would lead into a dynamic state where every new pre-processed burst triggers a processing update to detect unexpected changes from InSAR time series and therefore provides a signal for early warning against a potential dangerous displacement, e.g., a landslide, instability of an engineering structure or a formation of a sinkhole. An update of the processing chain would also allow use of cross-polarised Sentinel-1 data, needed for polarimetric analyses. The current system is running at a national supercomputing centre IT4Innovations in interconnection to the Czech Copernicus Collaborative Ground Segment (CESNET), providing fast on-demand InSAR results over Czech territories. A full nationwide PS processing using data over Czechia was performed in 2017, discovering several areas of land deformation. Its downsampled version and basic findings are demonstrated within the article., This work was supported by The Ministry of Education, Youth and Sports from the National Programme of Sustainability (NPU II) project “IT4Innovations excellence in science—LQ1602” and from the Large Infrastructures for Research, Experimental Development and Innovations project “IT4Innovations National Supercomputing Center—LM2015070”.

In the LEXIS project, a tsunami and earthquake large scale pilot is being deployed on top of an innovative HPC/Cloud orchestration layer, combining real-time deadlines, high performance simulations and cloud processing. To handle this, we have relied on a suitable model of computation oriented towards performance and real-time requirements, and used it to model, map and integrate the pilot workflow over the LEXIS orchestration and data exchanges technologies. This paper describes how the model of computation expresses real-time constraints over workflows, how such a model can be used to compute the necessary resources for a target latency and quality of service, and how the LEXIS orchestration layer is extended to offer the necessary features to implement this workflow.

Artificial intelligence (AI) is undergoing a revolution thanks to the breakthroughs of machine learning algorithms in computer vision, speech recognition, natural language processing and generative modelling. Recent works on publicly available pharmaceutical data showed that AI methods are highly promising for Drug Target prediction. However, the quality of public data might be different than that of industry data due to different labs reporting measurements, different measurement techniques, fewer samples and less diverse and specialized assays. As part of a European funded project (ExCAPE), that brought together expertise from pharmaceutical industry, machine learning, and high-performance computing, we investigated how well machine learning models obtained from public data can be transferred to internal pharmaceutical industry data. Our results show that machine learning models trained on public data can indeed maintain their predictive power to a large degree when applied to industry data. Moreover, we observed that deep learning derived machine learning models outperformed comparable models, which were trained by other machine learning algorithms, when applied to internal pharmaceutical company datasets. To our knowledge, this is the first large-scale study evaluating the potential of machine learning and especially deep learning directly at the level of industry-scale settings and moreover investigating the transferability of publicly learned target prediction models towards industrial bioactivity prediction pipelines.

Accurate and rapid earthquake loss assessments and tsunami early warnings are critical in modern society to allow for appropriate and timely emergency response decisions. In the LEXIS project, we seek to enhance the workflow of rapid loss assessments and emergency decision support systems by leveraging an orchestrated heterogeneous environment combining high-performance computing resources and Cloud infrastructure. The workflow consists of three main applications: firstly, after an earthquake occurs, its shaking distribution (ShakeMap) is computed based on the OpenQuake code. Secondly, if a tsunami may have been triggered by the earthquake, tsunami simulations (first a fast and coarse and later a high-resolution and computationally intensive analysis) are performed based on the TsunAWI simulation code that allows for an early warning in potentially affected areas. Finally, based on the previous results, a loss assessment based on a dynamic exposure model using open data such as OpenStreetMap is performed. To consolidate the workflow and ensure respect of the time constraints, we are developing an extension of a time-constrained dataflow model of computation, layered above and below the workflow management tools of both the high-performance computing resources and the Cloud infrastructure. This model of computation is also used to express tasks in the workflow at the right granularity to benefit from the data management optimisation facilities of the LEXIS project. This paper describes the workflow, the associated computations and the model of computation within the LEXIS platform.

Our automatized interferometric monitoring system, IT4S1, contains a database of Sentinel-1 satellite image bursts that have been preprocessed to the state of a consistent well-coregistered dataset. The coregistration solution introduces a new type of data, an SLC-C (corrected single look complex data). These are SLC images ready for interferometric analyses with maximally reduced phase signature of topography and flat-Earth phase. The further processing time for multitemporal interferometry techniques is significantly reduced for achieving velocity maps based on e.g. Persistent Scatterers or Small Baseline techniques. In this paper, authors focus on quality of estimated topographic phase, finding out that computation of perpendicular baseline between satellite positions based on precise orbit ephemerides can differ and thus the approach may suffer from a residual topographic signature or a flat-Earth phase. The paper presents conclusions taken by an investigation of one such pre-prepared dataset over a mountainous area in southern Spain. Though some orbital inparallelity occurs during Sentinel-1 acquisitions (causing inaccuracies of perpendicular baselines within first meters), the flat-Earth is removed properly using this approach. The paper further demonstrates results using SLC-C approach using available and custom processing techniques and notifies about the possibility that in terms of overall accuracy and precision of multitemporal interferometry this approach contains inaccuracies that can be either estimated as part of error phase screens or even neglected.

The probabilistic time-dependent vehicle routing problem is presented in this paper. It is a novel variant of the vehicle routing problem. The variant is a problem of finding optimal routes for a fleet of vehicles visiting customers in order to proceed delivery or pick-up. All customers must be visited in designated times with given probabilities and services must begin within the given time windows. Each vehicle has a given capacity, which must not be exceeded. The travel time between each two locations is dependent on a vehicle departure time. The paper also contains examples and solution results of testing data sets using the adaptive large neighbourhood search algorithm.

Floating car data (FCD) are one of the most important sources of traffic data. However, their processing requires several steps that may seem trivial but have far-reaching consequences. One such step is map-matching, i.e. assignment of the FCD measurement to the correct road segment. While it can be done very simply by assigning the point of measurement to the closest road, this approach may produce a highly undesirable level of error. The second challenge connected with processing of FCD measurements is missing measurements. They are usually caused by the shortcomings of GPS technology (e.g. the satellites can be obscured by buildings or bridges) and may deny us many measurements during longer downtimes. The last problem we will solve is the assignment of measurements to very short segments. FCD measurements are taken in periodic steps for several seconds long. However, some road segments are very short and can be passed by a car in the shorter interval. Such segments are therefore very difficult to monitor. We plan to solve all these problems through a combination of geometric map-matching with the Dijkstra’s shortest path algorithm.

In this paper we propose an innovative routing algorithm which takes into account stochastic properties of the road segments using the Probabilistic Time-Dependent Routing (PTDR). It can provide optimal routes for vehicles driving in a smart city based on a global view of the road network. We have implemented the algorithm in a distributed on-line service which can leverage heterogeneous resources such as Cloud or High Performance Computing (HPC) in order to serve a large number of clients simultaneously and efficiently. A preliminary experimental results using a custom traffic simulator are presented.

Effective navigation and distribution of traffic flow in large cities has become a hot topic in recent years. The authors have developed an advanced server side routing system which, together with client side navigation systems, is able not only to navigate cars according to their routing requests, but also to distribute traffic flow within a city. The main goal of the paper is to propose a distributed environment used for an advanced server side navigation system with a focus on effective usage of computational resources for different tasks that need to be solved within the system. A combination of cloud and high performance computing resources in one environment is proposed. The authors also developed a simulator for testing these distributed computational resources. The system and the simulator were tested on the infrastructure at IT4Innovations National Supercomputing Center in the Czech Republic.

The HPC-as-a-Service concept is to provide users with simple and intuitive access to a supercomputing infrastructure without the need to buy and manage their own physical servers or data centers. This article presents the commonly used services and implementations of this concept and introduces our own in-house application framework called High-End Application Execution Middleware (HEAppE Middleware). HEAppE’s universally designed software architecture enables unified access to different HPC systems through simple object-oriented web-based APIs, thus providing HPC capabilities to users without the necessity to manage the running jobs forms the command-line interface of the HPC scheduler directly on the cluster. This article also contains the list of several pilot use-cases from a number of thematic domains where the HEAppE Platform was successfully used. Two of those pilots, focusing on satellite image analysis and bioimage informatics, are presented in more detail.

Variety of real-life structures can be simplified by a graph. Such simplification emphasizes the structure represented by vertices connected via edges. A common method for the analysis of the vertices importance in a network is betweenness centrality. The centrality is computed using the information about the shortest paths that exist in a graph. This approach puts the importance on the edges that connect the vertices. However, not all vertices are equal. Some of them might be more important than others or have more significant influence on the behavior of the network. Therefore, we introduce the modification of the betweenness centrality algorithm that takes into account the vertex importance. This approach allows the further refinement of the betweenness centrality score to fulfill the needs of the network better. We show this idea on an example of the real traffic network. We test the performance of the algorithm on the traffic network data from the city of Bratislava, Slovakia to prove that the inclusion of the modification does not hinder the original algorithm much. We also provide a visualization of the traffic network of the city of Ostrava, the Czech Republic to show the effect of the vertex importance adjustment. The algorithm was parallelized by MPI (http://www.mpi-forum.org/) and was tested on the supercomputer Salomon (https://docs.it4i.cz/) at IT4Innovations National Supercomputing Center, the Czech Republic. 808 72 61

The quality of an opt imal solut ion of the Vehicle Rout ing Problem is st rongly depended on the sett ing of the configurat ion parameters of the algorithm. The paper is focused on the int roduct ion of hyperparameter search for solving the Vehicle Rout ing Problem using a HyperLoom plat form for defining and execut ing scient ific pipelines in a dist ributed environment . To give a concrete example, we focused on Periodic Vehicle Rout ing Problem for the waste collect ion. HyperLoom plat form was used to define and execute the hyperparameters sweep pipeline. The heurist ic algorithm was tested on a real benchmark of the waste collect ion in Ostrava, Czech Republic. The aim of our ca se was to effect ively combine the minimizat ion of the total t ravelled distance and the opt imizat ion of the fairness of the routes in terms of the standard deviat ion of a tour length. The waste collect ion problem was very extensive and computat ionally demanding, so it was necessary to use high performance comput ing architecture for test ing a large number of different sett ings of configurat ion parameters. The experiments were run on the supercomputer Salomon operated by IT4Innovations Nat ional Supercomput ing Center in the Czech Republic.

Analysis of the relationship between a large number of sequences is a significant problem in many different applications such as business processes, sport, voting, weblogs, etc. Generally, studying relationship is based on clustering the sequences and creating a network of relationships. Interpretation and validation of such results require a domain expert knowledge. In this paper, we propose a methodology which is able to provide an insight into the sequence dataset prior to the analysis and independently of a domain expert. Such information may be used to direct the analysis, identify sequences of interest and expose special patterns in the sequences. This methodology leverages tools such as transition matrix, Shannon entropy, complexity index, pairwise state occurrence, etc. Due to the low computational complexity of these methods, this approach is possible to use on the large datasets and help to identify the subsets of such datasets which should be inspected closer with more sophisticated tools. Ability to extract relevant information using the aforementioned tools was validated on two datasets, one from business processes simulation and the other from robot soccer game simulation.

The robot soccer game introduces a variable and dynamic environment for cooperating agents. Coverage of areas such as multi-agent systems, robot control, optimal path planning, real-time image processing and machine learning makes this domain very attractive. This article presents our approach to strategy description of the robot soccer game and a method of real-time strategy adaptation performed during the game. The real-time strategy adaptation method improves the strategy by adding new rules to it. During this process many new rules can be added to the original strategy, thus making it more robust but more difficult to manage. Therefore, this article presents our method for strategy reduction using representatives, in terms of the number of rules within the strategy, while preserving the quality of the adapted strategy. Strategy, as we defined it, describes a space from the real world in which we know the physical coordinates of objects located in it. Therefore, the methods we developed for strategy planning can be applied to it.

International audience; The ANTAREX project relies on a Domain Specific Language (DSL) based on Aspect Oriented Programming (AOP) concepts to allow applications to enforce extra functional properties such as energy-efficiency and performance and to optimize Quality of Service (QoS) in an adaptive way. The DSL approach allows the definition of energy-efficiency, performance, and adaptivity strategies as well as their enforcement at runtime through application autotuning and resource and power management. In this paper, we present an overview of the ANTAREX DSL and some of its capabilities through a number of examples, including how the DSL is applied in the context of one of the project use cases.

Waste companies need to reduce the cost of collection of the municipal waste, to increase the separation rate of different types of waste, or site of waste source. The collection of waste is an important logistic activity within any city. In this paper, we mainly focus on the daily commercial waste collection problem. One of the approaches for how to resolve this problem is to use optimization algorithms. Ant colony optimisation metaheuristic algorithm (ACO) was used to solve the problem in this paper. This algorithm was adapted for a real data set (Waste Collection). The aim of this paper is to adapt the ACO algorithm and run it on HPC infrastructure to resolve the waste collection problem. We used High-End Application Execution Middleware (HEAppE), that provides smart access to the supercomputing infrastructure (in our case Salomon cluster operated by IT4Innovations National Supercomputing Centre in the Czech Republic). The results showed that the paralelisation of the algorithm is beneficial and brings together with the supercomputing power the possibility to solve larger problems of this type.

225 218

International audience; Designing and optimizing applications for energy-efficient High Performance Computing systems up to the Exascale era is an extremely challenging problem. This paper presents the toolbox developed in the ANTAREX European project for autotuning and adaptivity in energy efficient HPC systems. In particular, the modules of the ANTAREX toolbox are described as well as some preliminary results of the application to two target use cases.

International audience; We introduce Kara1k, a new musical dataset composed of 2,000 analyzed songs thanks to a partnership with a karaoke company. The dataset is divided into 1,000 cover songs provided by Recisio Karafun application, and the corresponding 1,000 songs by the original artists. Kara1k is mainly dedicated toward cover song identification and singing voice analysis. For both tasks, it offers novel approaches, as each cover song is a studio-recorded song with the same arrangement as the original recording, but with different singers and musicians. Essentia, harmony-analyser, Marsyas, Vamp plugins and YAAFE have been used to extract audio features for each track in Kara1k. We provide metadata such as the title, genre, original artist, year, International Standard Recording Code and the ground truths for the singer's gender, backing vocals, duets and lyrics' language. Additionally, we provide the instrumental track and the pure singing voice track for each cover song. We showcase two use-case experiments for Kara1k. In the cover song identification task using the Dynamic Time Warping method, we provide a comparison of traditional and new features: chroma and MFCC features, chords and keys, and chroma and chord distances. We obtain 84-89% identification accuracy for three of the features, which justifies our focus on karaoke songs. In the supporting experiment on singer gender classification, we evaluate the difference in the performance in two conditions - a pure singing voice and the singing voice mixed with the background music. The Kara1k dataset is freely available under the KaraMIR project website.

This article describes statistical evaluation of the computational model for precipitation forecast and proposes a method for uncertainty modelling of rainfall-runoff models in the Floreon+ system based on this evaluation. The Monte-Carlo simulation method is used for estimating possible river discharge and provides several confidence intervals that can support the decisions in operational disaster management. Experiments with other parameters of the model and their influence on final river discharge are also discussed.

The main goal of this article is to describe the overview of Floreon+ system, an online flood monitoring and prediction system, which was primarily developed for the Moravian-Silesian region in the Czech Republic. Moreover, the article specifies the basic processes, which are implemented for running automatic and on-demand simulations that utilize the High Performance Computing (HPC) infrastructure. The main purpose of hydrodynamic models in the disaster management context is to provide an accurate overview of hydrologic situation in a given river catchment. In the event of extreme weather conditions, such as unusually heavy rainfall, these models could provide valuable information about imminent flood risk endangering a particular area. In the disaster management context, time plays a very significant role. Up to date and accurate results obtained in a short time can be very helpful. The availability of such results can be significantly improved by utilization of HPC resources and tools. The article describes the individual parts of the system in terms of data types, dynamic data processing, visualization, and the overall architecture.

We have developed HyperLoom - a platform for defining and executing scientific workflows in large-scale HPC systems. The computational tasks in such workflows often have non-trivial dependency patterns, unknown execution time and unknown sizes of generated outputs. HyperLoom enables to efficiently execute the workflows respecting task requirements and cluster resources agnostically to the shape or size of the workflow. Although HPC infrastructures provide an unbeatable performance, they may be unavailable or too expensive especially for small to medium workloads. Moreover, for some workloads, due to HPCs not very flexible resource allocation policy, the system energy efficiency may not be optimal at some stages of the execution. In contrast, current public cloud providers such as Amazon, Google or Exoscale allow users a comfortable and elastic way of deploying, scaling and disposing a virtualized cluster of almost any size. In this paper, we describe HyperLoom virtualization and evaluate its performance in a virtualized environment using workflows of various shapes and sizes. Finally, we discuss the Hyperloom potential for its expansion to cloud environments. 611 406 397