The 2023 International Conference on Computational Science and Its Applications
- ICCSA 2023 -

The National Technical University of Athens and University of the Aegean, Lesvos Island, Greece

July 3-6, 2023

The ICCSA 2023 conference will be organized on Lesvos Island, Greece. Virtual registration (with a reduced registration fee) is still possible but in the spirit of facilitating researchers suffering from very critical economic and political situations. Besides online sessions, there are pure on-site sessions with presential participation.

The ICCSA Conference offers an opportunity to bring together scientists of different disciplines, discuss new issues, tackle complex problems and find advanced solutions breeding new trends in Computational Science.

Among the workshops suggested during the conference:

*** Advanced Data Science Techniques with applications in Industry and Environmental Sustainability ***

This session is dedicated to advanced data science and machine learning techniques. These techniques can be applied in a large set of case studies (for example industrial applications and/or sustainability assessment).
Among the new trends in data science, this session focuses on several main topics:
- Explainability: machine learning algorithms can produce high performing results; however, the reliability and the industrialization of these algorithms require a comprehension of the underlying phenomena and patterns in order to make confident strategic decisions and, recently, to comply with new personal data protection regulations (like GDPR).

- Causality: traditional statistics and machine learning techniques are effective in getting the correlation between variables. Unfortunately, correlation is a weak starting point for decisions, while identifying causality relationships between variables can improve decision-making processes in industrial production or policy making.

- Data Fusion & Physic-informed Machine Learning: data is the current source of information on which, currently, most machine learning approaches rely on to fit the model (i.e., minimize the error). However, data is not the only source of knowledge. For example, in many contexts physics equations or regulations are available. The session welcomes applications of techniques that combine data with different sources of information.

- Data Science project management: training a model is only a part of the effort of a data science project. Before model training, it is necessary to properly collect data, clean the dataset, and perform data exploratory analysis. After the model is trained, it is necessary to monitor the performance against potential drifts. The session welcomes papers that propose solutions to help data scientists to deal with these kinds of challenges (outlier detection, drift monitoring, advanced exploratory data analysis, and advanced hypothesis testing).

- Process mining: process mining consists in the analysis of operational processes based on event logs. This is a relatively new field of data science that can help to understand processes, generate failure trees and ultimately make more informed decisions.

- Data augmentation: In real-world use cases, data collection is frequently difficult and expensive. Generating synthetic data points from real data is a common method for augmenting a dataset. However, the synthetic samples frequently propagate biases and are challenging to evaluate. This session welcomes papers presenting techniques for reliable data augmentation.

- From data (predictions) to context-aware decisions (prescriptions): The availability of data and advances in data science and machine learning give rise to novel forms of industrial practices in managing quantitative operations (e.g., enhanced predict-then-optimize frameworks, data-driven optimization). This session welcomes contributions that can help decision-makers to use data efficiently and effectively in order to support the decision-making process in providing context-aware and competitive solutions.

These are the technical topics we intend to discuss in this conference session. To be accepted, a contribution must be related to at least one of the topics listed above. We accept both theoretical and applied contributions. The list is not strictly exhaustive. Contributions dealing with similar topics may be accepted, but in case of doubt on the fit, it is suggested to ask the program committee before submission.

Application domains:
The data science topics are those discussed in the description. Their application domain is large and can include for example (the list is not exhaustive):
- Industrial process optimization;
- Environmental applications: water and energy saving;
- Decarbonisation;
- Energy consumption for machine learning application;
- Machine Learning & Material characterization;
- Technology assessment;
- Sensors and Internet of Things;
- Environmental pollution (air, soil, water, noise) control and assessment;
- Hazards and Risks assessment;
- Renewable resources (sustainability indicators, energy harvesting, grid integration);
- Resource optimization;
- Social data mining application to inform product optimization.


Submitted papers will be subject to stringent peer review by at least three experts and carefully evaluated based on originality, significance, technical soundness, and clarity of exposition. Accepted papers will appear in the Conference proceedings to be published by Springer-Verlag in Lecture Notes in Computer Science. The authors of a selected number of top-quality papers will be requested to extend their papers for further review and publication as special issues in highly recognized international journals.

ICCSA 2023 accepts paper submissions in the following categories: General Tracks and Workshops' Full Papers, General Tracks and Workshops' Short Papers and PhD Showcase Papers. The details for each paper submission category can be found at the URL https://www.iccsa.org/Instructions-for-authors and the list of Workshops at the URL: https://www.iccsa.org/workshops.