loading
Papers Papers/2022 Papers Papers/2022

Research.Publish.Connect.

Affiliation

Charles University, Czech Republic
Context (please use complete words only)
Field
Sort by

21 Papers

A Model-Based Approach to Experiment-Driven Evolution of ML Workflows

Topics: Modeling for AI Applications; Model-Based Software Development

Authors: Petr Hnětynka 1 , Tomáš Bureš 1 , Ilias Gerostathopoulos 2 , Milad Abdullah 1 and Keerthiga Rajenthiram 2

Affiliations: 1 Charles Univesity, Prague, Czech Republic , 2 Vrije Universiteit Amsterdam, The Netherlands

Digital Object Identifier: 10.5220/0013380500003896

Proceedings of MODELSWARD 2025, ISBN: 978-989-758-729-0

Collapse Button Quick Abstract
Machine Learning (ML) has advanced significantly, yet the development of ML workflows still relies heavily on expert intuition, limiting standardization. MLOps integrates ML workflows for reliability, while AutoML automates tasks like hyperparameter tuning. However, these approaches often overlook the iterative and experimental nature of the development of ML workflows. Within the ongoing ExtremeXP project (Horizon Europe), we propose an experiment-driven approach where systematic experimentation becomes central to ML workflow evolution. The framework created within the project supports transparent, reproducible, and adaptive experimentation through a formal metamodel and related domain-specific language. Key principles include traceable experiments for transparency, empowered decision-making for data scientists, and adaptive evolution through continuous feedback. In this paper, we present the framework from the model-based approach perspective. We discuss the lessons learned from the use of the metamodel-centric approach within the project—especially with use-case partners without prior modeling expertise.

Interpreting Workflow Architectures by LLMs

Topic: Pattern Recognition and Machine Learning for SSE

Authors: Michal Töpfer , Tomáš Bureš , František Plášil and Petr Hnětynka

Affiliation: Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic

Digital Object Identifier: 10.5220/0013358000003928

Proceedings of ENASE 2025, ISBN: 978-989-758-742-9

Collapse Button Quick Abstract
In this paper, we focus on how reliably can a Large Lanuage Model (LLM) interpret a software architecture, namely a workflow architecture (WA). Even though our initial experiments show that an LLM can answer specific questions about a WA, it is unclear how correct its answers are. To this end, we propose a methodology to assess whether an LLM can correctly interpret a WA specification. Based on the conjecture that the LLM needs to correctly answer low-abstraction level questions to answer questions at a higher abstraction level properly, we define a set of test patterns, each of them providing a template for low-abstraction level questions, together with a metric for evaluating the correctness of LLM’s answers. We posit that having this metric will allow us not only to establish which LLM model works the best with WAs, but also to determine what their concrete syntax and concepts are suitable to strengthen the correctness of LLM’s interpretability of WA specifications. We demonstrate the methodology on the workflow specification language developed for a currently running Horizon Europe project.

Modeling Machine Learning Concerns in Collective Adaptive Systems

Topics: Domain-Specific Modeling; Modeling for AI Applications; Model-Based Software Development

Authors: Petr Hnětynka , Martin Kruliš , Michal Töpfer and Tomáš Bureš

Affiliation: Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic

Digital Object Identifier: 10.5220/0011693300003402

Proceedings of MODELSWARD 2023, ISBN: 978-989-758-633-0

Collapse Button Quick Abstract
Collective adaptive systems (CAS) are systems composed of a large number of heterogeneous entities without central control that adapt their behavior to reach a common goal. Adaptation and collaboration in such systems are traditionally specified via a set of logical rules. Nevertheless, such rules are often too rigid and do not allow for the evolution of a system. Thus, recent approaches started with the introduction of machine learning (ML) methods into CAS. In the is paper, we present a model-driven approach showing how CAS, which employs ML methods for adaptation, can be modeled—on both the platform independent and specific levels. In particular, we define a meta-model for modeling CAS and a mapping of concepts defined in the meta-model to the Python framework.

Towards Model-driven Fuzzification of Adaptive Systems Specification

Topics: Model Transformation; Model Transformations and Generative Approaches; Modeling for AI Applications

Authors: Tomáš Bureš , Petr Hnětynka , Martin Kruliš and Jan Pacovský

Affiliation: Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic

Digital Object Identifier: 10.5220/0010910800003119

Proceedings of MODELSWARD 2022, ISBN: 978-989-758-550-0

Collapse Button Quick Abstract
The position paper outlines a method transforming adaptation rules in a self-adaptive system to a machine learning problem using neural networks. This makes it possible to endow a self-adaptive system with the possibility to learn. At the same time, by controlling the degree to which this transformation is done, one can scale the tradeoff between learning capacity and uncertainty in the self-adaptive system. The paper elaborates this process as a model transformation pipeline. The pipeline starts with a model capturing the strict adaptation rules. Then it is followed by multiple steps in which the strict rules are gradually fuzzified by well-defined transformations. The last model transformation in the pipeline transforms the fuzzified rules to a neural network that can be trained using the traditional stochastic gradient descent method. We briefly showcase this using two examples from the area of collective adaptive systems.

Using Procedure Cloning for Performance Optimization of Compiled Dynamic Languages

Topic: Programming Frameworks and Platforms

Authors: Robert Husák 1 , Jan Kofroň 1 , Jakub Míšek 2 and Filip Zavoral 1

Affiliations: 1 Faculty of Mathematics and Physics, Charles University, Ke Karlovu 3, Prague, Czech Republic , 2 iolevel s.r.o., Zelená 1743, Prague, Czech Republic

Digital Object Identifier: 10.5220/0011272300003266

Proceedings of ICSOFT 2022, ISBN: 978-989-758-588-3

Collapse Button Quick Abstract
Compilation of dynamic programming languages into strongly typed platforms such as .NET and JVM has proven useful in enhancing the security and interoperability of the resulting programs, as well as in enabling their source-less distribution. In order to produce the best intermediate code possible, dynamic language compilers can employ advanced interprocedural type analysis and perform various optimizing transformations. However, these efforts are often limited due to the ambiguity of types in these languages and the need to preserve soundness. In this paper, we improve the performance of global functions by adapting the technique of procedure cloning, focusing on different parameter types those specialized clones can be called with. We devise several heuristics to identify the most promising specializations and call them from their respective call sites. Our technique was implemented into PeachPie, a compiler of PHP to .NET, and evaluated on standard benchmarks distributed with PHP. Benchmarks containing deep recursion show a speedup factor up to 3.65, while benchmarks of computation-intensive loops reveal a speedup factor up to 2.64.

Increasing Business Opportunities for Drone Services

Topics: B2C/B2B Considerations; Service-oriented Enterprise ; Industrial/Manufacturing

Authors: S. Brezani 1 , R. Hrasko 1 , D. Vanco 1 and P. Vojtas 2 ; 1

Affiliations: 1 Globesy ltd., Framborska 58, Zilina, Slovak Republic , 2 Dpt. Software Engineering, Charles University, Malostranke nam. 25, Prague, Czech Republic

Digital Object Identifier: 10.5220/0010566900930100

Proceedings of ICE-B 2021, ISBN: 978-989-758-527-2

Collapse Button Quick Abstract
We describe our UAV imaging research and development from a business perspective. Our main goal is to increase business opportunities for our drone services. We build on experiences from an already deployed drone service provided by our IT company. The application is using a commercial tool for processing photogrammetry. Its advantage is accuracy, but the main disadvantage is the time needed for annotation by a trained human operator. Our methodology is based on user studies and knowledge gained in communicating with potential customers at IT trade fairs and exhibitions. We analyse the duration and automation of the service as key factors. We consider two types of higher automation of the solution. First is the automation of annotation - less accurate, without human intervention. The second is automation in flight planning and implementation. The use of other drone peripherals or hybrid drones can also create new types of services. In particular, there is a demand for immediate execution of on-site flight imaging without any pre-calibration. Our considerations, expanding our services, also include various inspections or direct involvement in industrial processes. Some improvements were tested on an experimental prototype. The results indicate improvement making services cheaper and faster.

Automatic Train Operation: History and Open Questions

Topics: Intelligent Infrastructure and Guidance Systems; Information Systems and Technologies

Authors: Aleš Lieskovský 1 , Ivo Myslivec 1 and Michal Žemlička 2

Affiliations: 1 Závod Technika, A ŽD Praha, s.r.o., Žirovnická 3146/2, Praha 10, Czech Republic , 2 Závod Technika, A ŽD Praha, s.r.o., Žirovnická 3146/2, Praha 10, Czech Republic, Department of Informatics, Jan Evangelista Purkyně University, České mládeže 8, Ústí nad Labem, Czech Republic, Department of Software and Computer Science Education, Charles University, Praha, Czech Republic

Digital Object Identifier: 10.5220/0009418002600267

Proceedings of VEHITS 2020, ISBN: 978-989-758-419-0

Collapse Button Quick Abstract
The paper presents the concept of automatic train operation. We give here short description of its functionality and remember some points from its history. There is an overview of various future development as well as proposals for improvement of some existing, especially mainline, solutions. There are presented also some observations from decades of practical use of automatic train operation in the Czech Republic. Selected challenges are presented and discussed.

Indexing Patterns in Graph Databases

Topics: Nosql Databases; Database Architecture and Performance

Authors: Jaroslav Pokorný 1 , Michal Valenta 2 and Martin Troup 2

Affiliations: 1 Faculty of Mathematics and Physics, Charles University, Prague and Czech Republic , 2 Faculty of Information Technology, Czech Technical University, Prague and Czech Republic

Digital Object Identifier: 10.5220/0006826903130321

Proceedings of DATA 2018, ISBN: 978-989-758-318-6

Collapse Button Quick Abstract
Nowadays graphs have become very popular in domains like social media analytics, healthcare, natural sciences, BI, networking, graph-based bibliographic IR, etc. Graph databases (GDB) allow simple and rapid retrieval of complex graph structures that are difficult to model in traditional IS based on a relational DBMS. GDB are designed to exploit relationships in data, which means they can uncover patterns difficult to detect using traditional methods. We introduce a new method for indexing graph patterns within a GDB modelled as a labelled property graph. The index is organized in a tree structure and stored in the same database where the database graph. The method is analysed and implemented for Neo4j GDB engine. It enables to create, use and update indexes that are used to speed-up the process of matching graph patterns. The paper provides a comparison between queries with and without using indexes.
Collapse Button Related Ontology Subjects/Areas/Topics (4)
Data Engineering

Databases and Data Security

Nosql Databases
Database Architecture and Performance

Automated Dynamic Formation of Component Ensembles - Taking Advantage of Component Cooperation Locality

Topic: Component-Based Software Engineering

Authors: Filip Krijt , Zbynek Jiracek , Tomas Bures , Petr Hnetynka and Frantisek Plasil

Affiliation: Charles University, Czech Republic

Digital Object Identifier: 10.5220/0006273705610568

Proceedings of MODELSWARD 2017, ISBN: 978-989-758-210-3

Collapse Button Quick Abstract
Smart cyber-physical systems (sCPS) is a growing research field focused on scenarios where a set of autonomous software-hardware entities (components) is cooperating via network communication to achieve a type of swarm or cloud intelligence. Typically the components’ cooperation is designed at a low level of abstraction and their behavior validated via simulations. As a remedy, a declarative language capable of specifying high-level component ensembles has been proposed in recent work. By capturing component functionality and the cooperation constraints, a specification serves both for generating platform-specific implementation and as a model@run.time to support self-adaption via dynamic formation of ensembles. However, for a particular specification, multiple possible architectural configurations exist with various impact on the system. Given their typically large number, we select the best one via an SMT solver. In this paper, we show that scalability of such approach can be supported by exploiting the effect of locality in component cooperation and by hoisting specific domain knowledge to the level of architecture.
Collapse Button Related Ontology Subjects/Areas/Topics (4)
Applications and Software Development

Model-Driven Software Development

Software Engineering
Component-Based Software Engineering

Data Preprocessing of eSport Game Records - Counter-Strike: Global Offensive

Topics: Information Quality; Information Integration; Data Integrity; Organizational Concepts and Practices; Data Management for Analytics

Authors: David Bednárek , Martin Krulis , Jakub Yaghob and Filip Zavoral

Affiliation: Charles University, Czech Republic

Digital Object Identifier: 10.5220/0006475002690276

Proceedings of DATA 2017, ISBN: 978-989-758-255-4

Collapse Button Quick Abstract
Electronic sports or pro gaming have become very popular in this millenium and the increased value of this new industry is attracting investors with various interests. One of these interest is game betting, which requires player and team rating, game result predictions, and fraud detection techniques. In our work, we focus on preprocessing data of Counter-Strike: Global Offensive game in order to employ subsequent data analysis methods for quantifying player performance. The data preprocessing is difficult since the data format is complex and undocumented, the data quality of available sources is low, and there is no direct way how to match players from the recorded files with players listed on public boards such as HLTV website. We have summarized our experience from the data preprocessing and provide a way how to establish a player matching based on their metadata.
Collapse Button Related Ontology Subjects/Areas/Topics (18)
Artificial Intelligence

Databases and Data Security

Integration/Interoperability

Collaboration and e-Services

e-Business

Knowledge Engineering and Ontology Development

Data Engineering

Enterprise Information Systems

Knowledge-Based Systems

Data Integrity

Information and Systems Security

Ontologies and the Semantic Web

Data Management and Quality

Information Integration

Organizational Concepts and Best Practices

Data Management for Analytics

Information Quality

Symbolic Systems

Topic(s):

Filter listed results, by

Publication date

From:
To:
Author
Locate Author in list below
Data pager
Data pager
12345
 Page 1 of 5, items 1 to 10 of 48.
Add Author(s) to filter:
Conference
Locate Conference in list below
Add Conference(s) to filter:

- Science and Technology Publications, Lda.
RESOURCES

Proceedings

Papers

Authors

Ontology

CONTACTS

Science and Technology Publications, Lda
Avenida de S. Francisco Xavier, Lote 7 Cv. C,
2900-616 Setúbal, Portugal.

Phone: +351 265 520 185 (National fixed network call)
Fax: +351 265 520 186
Email: info@scitepress.org

EXTERNAL LINKS

PRIMORIS

INSTICC

SCITEVENTS

CROSSREF

PROCEEDINGS SUBMITTED FOR INDEXATION BY:

dblp

Ei Compendex

SCOPUS

Semantic Scholar

Google Scholar

Microsoft Academic