A5027172970- Scientific Computing and Data Management
- Distributed and Parallel Computing Systems
- Research Data Management Practices
- Seismic Imaging and Inversion Techniques
- Peer-to-Peer Network Technologies
- Semantic Web and Ontologies
- Reservoir Engineering and Simulation Methods
- Geological Modeling and Analysis
- Topic Modeling
- Caching and Content Delivery
- Domain Adaptation and Few-Shot Learning
- Web Data Mining and Analysis
- Advanced Data Storage Technologies
- Big Data and Business Intelligence
- Meteorological Phenomena and Simulations
- Seismology and Earthquake Studies
- Recommender Systems and Techniques
- Data Quality and Management
- Anomaly Detection Techniques and Applications
- Climate variability and models
- Cloud Computing and Resource Management
- Generative Adversarial Networks and Image Synthesis
- Machine Learning and Data Classification
- Information Retrieval and Search Behavior
- Advanced Database Systems and Queries
National Centre of Scientific Research "Demokritos"
2016-2024
Institute of Informatics & Telecommunications
2019
Fraunhofer Institute for Intelligent Analysis and Information Systems
2017
University of Edinburgh
2012-2015
University of Glasgow
2003-2012
The DeployAI project [1] designs and delivers a fully operational European AI-on-Demand Platform (AIoDP) to empower the industry with access cutting-edge AI technology, promote trustworthy, ethical, transparent solutions, focus on SMEs public sector. To achieve this, platform enables development deployment of solutions through following core solutions: (i) Builder [2], which allows assembling reusable modules into pipelines; (ii) seamless Cloud HPC infrastructures (e.g., MeluXina LUMI);...
Disaster scenarios play a crucial role in research and preparedness efforts, providing basis to derive valuable insights into potential future disaster evolvements impact. Scenarios are composed of events, which can be either hypothetical or derived from dedicated databases that track disasters have occurred the past (e.g., https://www.emdat.be/). By leveraging historical data such databases, researchers applied various statistical methods analyze events their complex dependencies [1]....
This paper presents dispel4py, a new Python framework for describing abstract stream-based workflows distributed data-intensive applications. The main aim of dispel4py is to enable scientists focus on their computation instead being distracted by details the computing infrastructure they use. Therefore, special care has been taken provide with ability map different enactment platforms dynamically, at run time. In this work we present four mappings: Apache Storm, MPI, multi-threading and...
Peer-to-Peer (P2P) networking is aimed at exploiting the potential of widely distributed information pools and its effortless access retrieval irrespectively underlying protocols, operating systems devices. However, prohibiting limitations have been identified perhaps most important one successful location relevant sources efficient query routing in large, highly P2P networks. In this paper, a novel, cluster-based architecture for IR over networks presented evaluation focused on...
In this paper we describe the design, and implementation of Open Science Data Cloud, or OSDC. The goal OSDC is to provide petabyte-scale data cloud infrastructure related services for scientists working with large quantities data. Currently, consists more than 2000 cores 2 PB storage distributed across four centers connected by 10G networks. We discuss some lessons learned during past three years operation software stacks used in also research projects biology, earth sciences, social sciences enabled
This paper presents dispel4py, a new Python framework for describing abstract stream-based workflows distributed data-intensive applications. These combine the familiarity of programming with scalability workflows. Data streaming is used to gain performance, rapid prototyping and applicability live observations. dispel4py enables scientists focus on their scientific goals, avoiding distracting details retaining flexibility over computing infrastructure they use. The implementation,...
We present dispel4py a versatile data-intensive kit presented as standard Python library.It empowers scientists to experiment and test ideas using their familiar rapid-prototyping environment.It delivers mappings diverse computing infrastructures, including cloud technologies, HPC architectures specialised machines, move seamlessly into production with large-scale data loads.The are fully automated, so that the encoded analyses handling completely unchanged.The underpinning model is...
Numerical Weather Prediction (NWP) simulations produce meteorological data in various spatial and temporal scales, depending on the application requirements. In current study, a deep learning approach, based convolutional autoencoders, is explored to effectively correct error of NWP simulation. An undercomplete autoencoder (CAE) applied as part dynamic correction data. This work an attempt improve seasonal forecast (3–6 months ahead) accuracy for Greece using global reanalysis dataset (that...
The VERCE project has pioneered an e-Infrastructure to support researchers using established simulation codes on high-performance computers in conjunction with multiple sources of observational data. This is accessed and organised via the science gateway that makes it convenient for seismologists use these resources from any location Internet. Their data handling made flexible scalable by two Python libraries, ObsPy dispel4py services delivered ORFEUS EUDAT. Provenance driven tools enable...
No abstract available.
The DARE platform has been designed to help research developers deliver user-facing applications and solutions over diverse underlying e-infrastructures, data computational contexts. is Cloud-ready, relies on the exposure of APIs, which are suitable for raising abstraction level hiding complexity. At its core, implements cataloguing execution fine-grained Python-based dispel4py workflows as services. Reflection achieved via a logical knowledge base, comprising multiple internal catalogues,...
The DARE platform enables researchers and their developers to exploit more capabilities handle complexity scale in data, computation collaboration. Today's challenges pose increasing urgent demands for this combination of capabilities. To meet technical, economic governance constraints, application communities must use shared digital infrastructure principally via virtualisation mapping. This requires precise abstractions that retain meaning while implementations infrastructures change....
The concept of FAIR data is being increasingly adopted at national and global levels as a way maximising the impact transparency publicly-funded research outcomes data. In this paper we introduce requirements Fusion community well technological directions proposed by Fair4Fusion project aiming increasing accessibility to fusion
This paper identifies the high value to researchers in many disciplines of having web-based graphical editors for scientific workflows and draws attention two technological transitions: good quality can now run a browser workflow enactment systems are emerging that manage multiple languages support multi-lingual workflows. We contend this provides unique opportunity introduce which turn would yield substantial benefits: users find it easier share combine methods encoded languages, common...