Professor Przemysław Kazienko: To be a significant player in the AI industry we must be visible and have more scientists and research projects

Herein below I present my own analysis and ideas connected with possibilities of faster development of artificial intelligence in Poland.

I. Key problems stemming development of AI in Poland

Problem 1: too few scientists and doctoral students from abroad

The most advanced research in artificial intelligence (and, more generally, in information technology) is conducted in the United States. The second place would probably go to the best universities of Western Europe, and the third – to some centers in Asia, mostly in Singapore and China. Poles are lagging behind the best teams of the above centers.

Our studies on efficiency of grants of the National Science Centre show that projects carried out in collaboration with foreign centers (National Science Centre Harmonia program) are (on average) the most efficient ones. A number of co authors from abroad involved in studies conducted in Poland is not very high either.

Problem 2: too few international research projects

Examples: Polish teams have acquired less than 1 percent of funds available within the H2020 program [Ed. note: EU Horizon 2020] although about 35 billion euros have already been earmarked.

  • As far as H2020 is concerned, information and computer technology (ICT) comes out best, in relative terms. A similar result was seen in the case of the FP7 program [Seventh Framework Program for research and technological development].
  • Polish teams have acquired only 27 ERC3 projects [ERC: European Research Council], mostly Starting Grants, with as many as 5 of them being in the field of information technology [PE6 – EU engineering sciences grant]. Polish information technology has the biggest number of ERC projects if compared to all other fields. In contrast, Switzerland has acquired 581 projects (33 in ICT) and Israel has acquired 482 (48 in ICT).
  • There are few ICT projects, including those connected with artificial intelligence, that are carried out for foreign entities (international companies or organizations, e.g. NATO).
  • To my knowledge there is not a single Polish team that would carry out an international research project in the field of artificial intelligence within the CHIST-ERA program (the program covers Analog Computing for Artificial Intelligence and Smart Distribution of Computing in Dynamic Networks – SDCDN).

Problem 3: hard to be noticed abroad

The achievements of Polish teams are often badly popularized – at conferences and in magazines addressed to a limited audience. In consequence, the number of citations of Polish studies is relatively low. Examples:

  • At the KDD conference (the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining) only 8 out of 3,102 published articles were written by authors having links with Poland. To compare, Czechs had 2 studies, Spaniards had 44, and Slovenians had 95.
  • At the NIPS conference (Conference on Neural Information Processing Systems) only 4 papers of Polish authors were published, with the overall number of studies reaching 7,261.
  • At the ICDM conference (The IEEE International Conference on Data Mining) 22 Polish papers were published although the total number of studies amounted to 3,129.

We do slightly better at conferences that are less renowned and recognized:

  • At the ECML PKDD conference (The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases – a worse, European editions of the KDD conference) 65 Polish articles were published out of 2,851 total.
  • At the ASONAM conference (The IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining) out of 1,615 published papers 31 were Polish.

We have good programmers of internet services, games or ERP systems but too few talents are encouraged to focus on more ambitious projects

The situation is not any better in the case of the most renowned scientific journals. Below you will find the list of professional journals with the highest impact factor (IF) in the field of artificial intelligence:

  • “Journal of Computer Vision” (IF=11.5): 2 papers from Poland out of 2,098 total (19 papers from the Czech Republic, 62 from Spain, 4 from Romania, 6 from Slovenia);
  • “IEEE Transactions on Pattern Analysis and Machine Intelligence” (IF=9.5): 7 papers from Poland out of 5,951 total;
  • “IEEE Transactions on Cybernetics” (IF=8,8): 14 papers from Poland out of 1,918 total;
  • “IEEE Transactions on Fuzzy Systems” (IF=8.4): 669 out of 2,563;
  • “IEEE Transactions on Evolutionary Computation” (IF=8.1): 27 out of 1,101;
  • “IEEE Transactions on Neural Networks and Learning Systems” (IF=8.0): 33 out of 1,957;
  • “Neural Networks” (IF=7.2): 29 out of 4,203;
  • “Information Fusion” (IF=6.6): 16 out of 925.

Sadly, a majority of the above publications, and often most of those published in journals, are written by members of Polish diaspora, i.e. scientists of Polish origin who work abroad but who also have links with Poland.

Problem 4: insufficient interdisciplinarity of research

Due to lack of tradition, mentality and lack of major system and support solutions, a substantial part of research in the field of artificial intelligence (and also ICT) is conducted only by teams of ICT specialists.

Problem 5: poor cooperation between centers

The vast majority of research is carried out in teams operating within one institution. Rarely do graduates decide to write their doctoral theses elsewhere than at a university where they were awarded their master’s degree. That makes the transfer of good practices and solutions impossible and undermines the effort to be involved in bigger industrial and research projects.

Problem 6: small critical mass; small number of doctoral theses in the field of artificial intelligence

Relatively few people (ICT specialists) focus on artificial intelligence. We have good programmers of internet services, games or ERP systems but too few talents are encouraged to focus on more ambitious projects, including artificial intelligence projects. Deficiencies in that field are reflected for example by a scarce number of doctoral theses submitted for competitions for the best doctoral thesis organized by the Polish Artificial Intelligence Society.

II. Proposed solutions and actions

The limitations I listed above may be overcome by implementing a policy that would stimulate suitable actions.

To increase the number of people (scientists) tasked with artificial intelligence in Poland:

  1. A dedicated national program for doctoral theses on artificial intelligence offering suitable scholarships (e.g. similar to those offered in the case of National Science Center projects, i.e. PLN 4,500 a month) granted in a national competition and obligating students to carry out a part of their works in another Polish center and to serve an internship abroad.
  2. An extensive program of scholarships/prizes for master’s theses in the field of artificial intelligence.
  3. Media programs to popularize artificial intelligence for a general audience.

To internationalize research, to improve its quality and visibility abroad:

  1. A program similar to CHIST-ERA for the states of the Visegrad Group (or Central Europe) dedicated to artificial intelligence. Financing: from the funds of the National Science Center and counterpart agencies in the states of the Group.
  2. An extensive program of prizes for Polish authors of publications on artificial intelligence (employed in Polish centers under an employment contract). The program would pertain only to the best international conferences and scientific journals.
  3. A program to employ scientists from abroad in the field of artificial intelligence, following a national competitive process. Addressed separately to young doctors (post-doc), separately to professors, and separately to scientists ready to spend their sabbatical leave in Poland. Additionally, there would be an obligation to stay for a short time in other Polish centers.
  4. Creation of Central-European Institute for Artificial Intelligence, which would both bring together best Polish teams and stay open to new groups. Funds for luring scientists to that center, including researchers from Central and Eastern Europe. Requirement – at least 50 percent of scientists from abroad. Financed from EU structural measures. The center might be responsible for providing support for other ideas mentioned in this text, observing all full transparency and openness rules. To ensure that such rules are followed, it would be recommended to appoint a suitable Board of Trustees, preferably consisting also of members from abroad.
  5. “European Centre for Computational Processing of Slavic Languages” association, i.e. the center for analysis and processing of Slavic languages.
  6. A program of cooperation with Polish scientists working in renowned foreign centers on artificial intelligence (also from industrial research centers, e.g. Google), including: a doctoral program obligating a scientist to return to Poland for a specific period of time (also the Double PhD Diploma program) and a visiting professors program.

To strengthen national cooperation, research interdisciplinarity and cooperation with industry (actions partially included in additional requirements in the above mentioned propositions):

  1. A program for interdisciplinary research projects in the field of artificial intelligence, e.g. dedicated Symfonia program (National Science Center) or TEAM-NET (Foundation for Polish Science).
  2. Dedicated programs for start ups (including BRidge Alfa, National Center for Research and Development) in the field of artificial intelligence.
  3. Support for platforms of cooperation between Polish industry and science, e.g. for SciCup platform used to organize open competitions for analysis of industrial data.

III. Areas of competence and research team

The team I lead is composed of about a dozen of people and works within ENGINE – The European Centre for Data Science, operating mainly within the Department of Computational Intelligence. We have proven our competences with publications in renowned journals, e.g. “Scientific Reports – Nature”, “Information Sciences” and with speeches and lectures given at most acclaimed international scientific conferences (KDD, ECML PKDD, ASONAM). The competences include the following fields connected with artificial intelligence:

  • analysis of social phenomena, e.g. diffusion of information, opinion dynamics, information manipulation, etc.;
  • analysis of social media and electronic media;
  • scientometrics and computational science of science, i.a. analysis of efficiency of sciences and research projects, identification of prospective research subject matters;
  • processing and analysis of social networks and complex networks; analysis of natural language texts;
  • preventive methods and algorithms for data science;
  • machine learning, namely for complex and dynamic data;
  • analysis of network data structures, also with the use of deep machine learning.

The team has completed several EU projects (FP7, H2020), several dozens of projects financed by the National Science Center, National Center for Research and Development, Ministry of Science and Higher Education, and also more than a dozen of projects financed by the industry. Most of projects were interdisciplinary and involved different sectors of economy (finance, telecommunication, production, commerce) and science (physics, social science, medicine, economics). The team is now working on several projects valued over PLN 30 million in total.

All team members have served or are serving research internships abroad, mainly at renowned universities in the USA (Stanford, UC, RPI, Notre Dame), Singapore (NTU) and Australia (UTS).

Team members include professor Nitesh Chawla from the University of Notre Dame (USA) working at our university on his own National Science Center project, and Suman Kundu, PhD, from India (who was awarded his doctoral degree at Indian Statistical Institute), a post-doc employed under an employment contract within a National Science Center project. In recent years the employees included more than a dozen of people from abroad financed by the EU FP7 project.

The team is strongly committed to a new master’s major: data science, which has a lot in common with artificial intelligence (e.g. machine learning), although it is not identical.

The article is based on a study by professor Przemysław Kazienko entitled “Development of artificial intelligence in Poland” (Wrocław 2018), prepared for the National Information Processing Institute.

Skip to content