Research areas of the ELLIS Unit Warsaw

The research activities of ELLIS Unit Warsaw are grouped into six areas: 

  1. The “Intelligent Algorithms and Learned Data Structures” area focuses on explainable algorithmic tools, learning data structures and algorithmic tools for Data Science and ML.
  2. The “Efficient and Sustainable Machine Learning focused on Computer Vision” area concentrates on efficient machine learning algorithms for computer vision, in particular on conditioning network computations, leveraging partial information for faster inference, and accumulating knowledge in continually trained models.
  3. The “Machine Learning and Sequential Decision-Making” area develops methods that can be applied in control to obtain broadly intelligent agents that could be deployed in the real world.
  4. The “Algorithmic Game Theory in Security” area develops multi-level management systems to protect critical infrastructure as well as systems for securing key state services against both kinetic and cyber threats.
  5. The “Interpretable Artificial Intelligence” area focuses on developing machine learning models that provide trustworthy and comprehensible explanations of how complex artificial intelligence models work.
  6. The “Autonomous Agents and Alignment of Language Models” area aims to develop training protocols that empower artificial agents to collect the most informative learning data and apply these protocols to align language models.

Publications

Visit our GitHub to check our publications.

Projects

ELIAS Project

IDEAS NCBR is implementing the ELIAS – European Lighthouse of AI for Sustainability project, which aims to establish Europe as a leader in artificial intelligence research.

The project has received funding from the European Union’s Horizon Europe Research and Innovation Programme under grant agreement No. 101120237 (ELIAS). Project duration: 01/09/2023 – 31/08/2027. Project value: 215,750 EUR.

The project is implemented by a consortium of leading European institutions from 17 countries. The scientific excellence of the consortium is guaranteed by the leading position of the members in their respective fields, as measured by their international visibility and impact.

The ELIAS initiative aims to create a Network of Excellence, linking researchers from academia with practitioners from industry, in order to distinguish Europe as a region where AI research contributes to a sustainable long-term future for our planet, a cohesive society and respect for individual preferences and rights.

At IDEAS NCBR this project is implementing by Prof. Tomasz Trzciński, Prof. Bartosz Zieliński and Dr. Bartłomiej Twardowski, members of the ELLIS Unit Warsaw.

For more information, please visit: elias-ai.eu

Project beneficiaries: Universita Degli Studi di Trento (lider of the ELIAS Project); Universiteit van Amsterdam; Politecnico di Milano; Università Degli Studi di Modena e Reggio Emilia; Institut Polytechnique de Paris; Eberhard Karls Universitaet Tuebingen; Fondazione Bruno Kessler; Fundación de la Comunitat Valenciana Unidad Ellis Alicante; Ethniko Kentro Erevnas Kai Technologikis Anaptyxis; Universitat de Valencia; Hasso-Plattner-Institut Für Digital Engineering GmbH; Università Degli Studi Di Milano; Kobenhavns Universitet; Institut Jozef Stefan; Ceske Vysoke Uceni Technicke V Praze; Universite De Toulouse; Max-Planck-Gesellschaft zur Forderung der Wissenschaften Ev; Umeå Universitet; Fondazione Istituto Italiano di Tecnologia; Universitatea Politehnica Din Bucuresti; IDEAS NCBR Sp z o.o.; Bitdefender Srl; IBM Ireland Limited; Institut National de Recherche en Informatique et Automatique; Robert Bosch GmbH; Robert Bosch Kft; Fondation De L’institut De Recherche Idiap; University of Manchester; Eidgenoessische Technische Hochschule Zuerich; Aalto Korkeakoulusaatio Sr.

Three projects from the National Science Centre at the ELLIS Unit Warsaw: from SONATA BIS 13, OPUS 26 and SONATA 19 calls 

Bartosz Zieliński, Łukasz Kuciński and Bartłomiej Twardowski received grants for a total of PLN 2,230,593 for their research projects from the Polish National Science Centre (NCN).

Bartosz Zieliński received PLN 990,600 from the SONATA BIS 13 call for the project ‘Sustainable computer vision for autonomous machines’. This project will adapt machine learning models to new data types and create architectures for limited computing resources, prioritizing efficiency and carbon footprint reduction. The developed solutions could be used in drones to protect national parks, including preventing animal poaching. They will enable fast, efficient monitoring of large remote areas, allowing for the tracking of animal movements and early detection of forest fires.

Łukasz Kuciński received PLN 624,640 from the OPUS 26 call for the project ‘Alignment of large language models via debate and reinforcement learning’, which aims to ensure large language models like GPT align with human values through debate-based training. Instead of using traditional methods, models are trained to argue and defend their points, engage with opposing views, and persuade judges of their stance. Employing a ‘teacher-student’ framework means AI ‘students’ will participate in debates overseen by a ‘teacher’. Outcomes will include a debate-based training protocol, public dataset of debate transcripts, trained model weights and source code, enhancing AI’s accuracy and decision-making.

Bartłomiej Twardowski received PLN 615,353 from the SONATA 19 call for the ‘MLAdapt – self-adapting machine learning’ project. The project focuses on challenges of deep learning in computer vision, aiming to create algorithms that enable continuous learning and adaptation, e.g. allowing autonomous vehicles to learn to navigate obstacles and move in difficult terrain. The models will use information available during deployment to improve their capabilities, similar to learning in humans. This approach promises applications in fields such as medicine, automotive industry, and robotics.

 

R-GRID Project

The main goal of the R-GRID project is to create an artificial intelligence tool to protect power grid systems. The project addresses one of the priorities for cooperation identified in 2023 by the NATO-Ukraine Joint Working Group on Scientific and Environmental Cooperation, which aims to provide solutions to Ukraine’s current and upcoming needs. The project consortium consists of: the Polish Association for National Security, the Ukrainian Institute for the Future, IDEAS NCBR and Laurea University of Applied Sciences from Finland.

R-GRID will use artificial intelligence to prevent power outages in key sectors or a complete blackout. Simulations generated by R-GRID are intended to help identify critical network elements and increase the resilience of energy systems − considering both traditional and renewable energy sources at various levels of technological advancement.

The R-GRID project is implemented by an international consortium managed by:

  • Maciej Kluczyński, NATO country Project Director, Polish Association for National Security,
  • Dr. Andrian Prokip, Partner country Project Director, Ukrainian Institute for the Future,
  • Dr. Tomasz Michalak, co-director, IDEAS NCBR Sp. z o. o., member of the ELLIS Unit Warsaw,
  • Dr. Päivi Mattila, co-director, Laurea University of Applied Sciences.

The project is planned to last 2 years and will open the way to the implementation and use of R-GRID in practice.

About the NATO Science for Peace and Security (SPS) Programme

The NATO SPS Programme promotes dialogue and practical cooperation between NATO nations and Partner countries based on scientific research, technological innovation and knowledge exchange. The SPS Programme offers funding, expert advice and support to tailor-made, civil security-relevant activities. SPS activities are guided by a set of Allied-approved key priorities that address emerging security challenges. SPS supports four types of activities: research and development Multi-Year Projects (MYP), Advanced Research Workshops (ARW), Advanced Training Courses (ATC), and Advanced Study Institutes (ASI). Its activities bring together academics, experts and officials from NATO and partner countries who jointly lead research and knowledge exchange activities. Opportunities for cooperation are announced as calls for proposals on the SPS website. 

Read more

EXALT Project

The grant for implementing the EXplainable ALgorithmic Tools (EXALT) project has been awarded by the European Research Council (ERC) and is funded through the European Union’s HORIZON ERC Proof of Concept Grants program. ERC Proof of Concept Grants are intended to facilitate the application of research findings into practical use and commercialization.

The EXALT project is led by Prof. Piotr Sankowski at IDEAS NCBR, who is a member of the ELLIS Unit Warsaw. The research conducted under the grant focuses on explainable artificial intelligence with the aim of better understanding algorithmic pathways. The research project aims to develop a tool that will enrich optimization algorithms with human-understandable explanations. The research will focus on simple optimization algorithms, for which an additional algorithm will be developed to determine why the original algorithm made a specific decision.

The results of the research conducted within the project may be useful in developing artificial intelligence-based solutions in areas such as medicine, customer service, or sales. If the investigated tool proves to be effective, its commercialization will commence.

Project Duration: September 1, 2023 – February 28, 2025
Grant Agreement Number: 101082299
Project Value: 150,000 EUR

Funded by the European Union. The views and opinions expressed are solely those of the author(s) and do not necessarily reflect the views of the European Union. Neither the European Union nor the funding body shall be liable for them.
 

Mazovia EDIH Project


Mazovia EDIH is one of the 11 European Digital Innovation Hubs (EDIH) in Poland. Its mission is to support micro, small, and medium-sized enterprises (SMEs) in strengthening their competitive position in the market through digital transformation.

Collaboration with Mazovia EDIH brings a range of benefits:

  • Support in adopting digital technologies – Through Mazovia EDIH, businesses can access the latest digital solutions, positively impacting the efficiency of their operations.
  • Opportunity to test new solutions before investment – Before full implementation, businesses can test and verify planned investments or projects, enabling informed investment decisions and minimizing the risk of loss.
  • Employee skills development – Mazovia EDIH offers training in digital technologies, thereby increasing the qualifications of employees and the innovation capacity of the enterprise.
  • Encouragement for continuous improvement – Companies are motivated to continuously improve their business and technological practices, leading to increased competitiveness in the market.

As a Partner, IDEAS NCBR’s task is to support SMEs in the areas of artificial intelligence and digital economics. At IDEAS NCBR this project is implementing by Prof. Tomasz Trzciński and Dr. Tomasz Michalak, members of the ELLIS Unit Warsaw.

Description of services offered within MAZOVIA EDIH:

1. Sustainable development in machine learning: Consultation, assessment, optimization, and implementation support for zero-waste techniques for companies

IDEAS NCBR offers services related to creating efficient machine learning models based on “zero-waste” techniques and sustainable development. Through our consultations, assessment, and optimization of machine learning models, companies can achieve higher performance and minimize resource consumption. Additionally, we provide training and workshops for companies, universities, and other organizations interested in machine learning based on “zero-waste” techniques.

Business benefits for entrepreneurs primarily include improved performance of machine learning models, cost reduction associated with resource consumption, and minimizing environmental impact through the use of “zero-waste” techniques.

2. Training: Increasing the efficiency of neural networks

The training is aimed at individuals involved in deep learning daily who want to stay up-to-date with the latest trends. In particular, the training is intended for programmers implementing neural networks in environments with limited computational resources, such as microcontrollers or mobile devices. The training covers various topics related to increasing the efficiency of artificial neural networks, especially those addressed within the Zero Waste Machine Learning project.

During the training, participants will learn algorithms for increasing the efficiency of neural networks and gain practical skills.

Get to know more: mazovia.edih.pl

The Mazovia EDIH project is co-financed by the European Commission under the Digital Europe Program for the years 2021-2027 (Grant Agreement with the EC No: 101083509 – Mazovia EDIH – DIGITAL-2021-EDIH-01) and co-financed with funds from the European Union under the Operational Program European Funds for the Modern Economy 2021–2027, action FENG.02.22 Co-financing of EDIH activities under Priority 2 Innovative Environment (Grant Agreement with PARP No: FENG.02.22-IP.02-0002/23-00). The Leader of the Mazovia EDIH Consortium is the Łukasiewicz Research Network – Industrial Research Institute for Automation and Measurements PIAP. Project implementation period: 01/01/2023 – 31/12/2025. Project value: 12,106,208.68 PLN. Amount of European Funds contribution: 12,106,208.58 PLN.