National Excellence Program for researching the mathematical foundations of artificial intelligence successfully completed
Under the leadership of the ELKH Alfréd Rényi Institute of Mathematics (Rényi), a large-scale National Excellence Program targeted at researching the mathematical foundations of artificial intelligence has been implemented. The program funded by the National Research, Development and Innovation Office (NKFIH) has achieved a number of outstanding results in terms of science, infrastructure development, education and innovation. The last ten years have seen revolutionary breakthroughs in artificial intelligence around the world. By funding the project, the NKFIH aimed to support Hungary’s efforts to regain lost ground in the field of machine learning, as it is of strategic importance that Hungary does not miss out on the economic, social and scientific benefits of these innovative results. The other main objective of the project was to create and develop knowledge centers where researchers from different areas of basic research are involved in artificial intelligence research.
In addition to Rényi, the ELKH Institute for Computer Science and Control (SZTAKI), Eötvös Loránd University (ELTE), Pázmány Péter Catholic University (PPKE) and the University of Szeged (SZTE) also played important roles in the implementation of the program’s objectives. The main results of the project were presented at a closing conference on 17 September 2021.
At the time of the Covid outbreak, the program was half-way to completion. This could have led to the research being interrupted, but just the opposite happened. In a difficult period, all participating institutions threw themselves into the project with renewed vigour and contributed to the results of the research teams already working on the topics in question. Significant progress has been made in representation learning, interpretability and statistical machine learning, as well as in several other areas.
Main research areas and findings of the consortium members
At Rényi, the project placed particular emphasis on research into the theoretical foundations of artificial intelligence. Of particular interest are, among other areas, the results of their research on the mathematical properties of information metrics and the detailed description of the structure of compressed information that can be extracted from large data sets using higher order Fourier analysis. They developed novel methods for comparing the internal states of artificial neural networks and have used this toolkit to investigate the properties of the learning process. They have also developed models for the control of automatic verification systems for theorems based on reinforcement learning. Based on these results, they have conducted combinatorial game theory research, investigated the smoothness behaviour of mappings implemented by artificial neural networks, and described loss function members that promote smoothness.
The research work carried out in a cooperation between Rényi and MedInnoScan Ltd focused on the practical application of artificial intelligence. As part of the program, a medical pilot application was developed to improve the treatment of around 200,000 patients with chronic wounds in Hungary. The treatment of these patients places a significant burden on the entire health care system, as they require continuous treatment, with dressings every two days on average. It makes sense to change the type of dressing depending on how the wound changes, and the artificial intelligence application embedded in the mobile app, the prototype of which was created by the researchers and presented at the closing conference, will help in this decision. The image database on which the research and AI training is based has been compiled through a national partnership. Led by the University of Pécs and with the active participation of nearly 70 medical institutions, four medical universities, the National Medical Rehabilitation Institute, a dozen hospitals and more than 50 nursing services and private wound care centers, 220,000 images of the chronic wounds of 5,500 patients were recorded. After clinical trials, the product is likely to be made available to professional nurses next year.
The main task of SZTAKI under the program was to carry out basic research motivated by practical needs, and to apply and demonstrate the results. Their research covered the information geometry foundations of deep learning, network theory, recommendation systems and reinforcement learning, and their applications spanned areas such as machine perception, social media analysis, robotics and autonomous transportation. During the three years of the program, a number of outstanding scientific papers were published and their results were presented at several public events. Key demonstrations: a portrait drawing robot; robotic applications, control and manufacture of drone flocks with reinforcement learning; a solution for exploring the inner functions of recommendation systems; detection of certain traffic objects; and an opinion analysis system for Covid vaccinations.
The ELTE Institute of Mathematics studied several areas of machine learning and their use in applied mathematics. In machine vision, the Institute investigated the teaching of deep neural networks and its specificities in several application domains (e.g. in the processing of medical images and satellite images). In the application of mathematical modelling, the efficiency of algorithms and procedures already used in practice was investigated. One such topic is the use of neural networks to support holistic programming technologies that play a central role in solving logistics or optimization problems in general. Another solution is to use neural networks to estimate parameters of stochastic processes, which are also used in the analysis of financial processes. In the context of the previous topics, research has extended to the implementation of deep learning models of natural language processing in other, new applications. The project has also had a significant impact on university education. Dozens of students have been involved in research, projects and theses in the inter- and transdisciplinary fields of artificial intelligence and mathematics. The launch of the “Mathematics Expert in Data Analytics and Machine Learning” course in English at the Institute is another long-term impact of the project.
The results of the PPKE Faculty of Information Technology and Bionics were presented at prestigious international forums and in peer-reviewed journals, including in the areas of wavelet-based segmentation and multi-discriminator GAN networks. The researchers also focused on proving the origins of neural networks and developing methods to prevent the theft of network weights. They also investigated the vulnerability of neural networks to attack and the reversibility of attacks. Their research in linguistic technology includes automatically generated vector space models of texts and the correction of incorrect, so-called noisy texts that deviate from the colloquial norm. The faculty’s achievements also include a patent filed under positive novelty research, the idea for which was motivated by the tragic sinking of the hungarian riverboat Hableány. The aim of the patent is to implement a machine learning system that makes the search and rescue of victims much easier and possible even in difficult circumstances.
The SZTE mainly investigated the interpretability and vulnerability of artificial intelligence algorithms. Both topics relate to the ‘black box’ problem of artificial intelligence. In natural language processing, interpretable and less resource-intensive report representations and multilingual models have been proposed, and research on the theoretical background of interpretability has been carried out. The researchers identified unprecedented vulnerabilities in the sensitivity of artificial intelligence algorithms to formal verification and have also investigated the limitations of simultaneous attacks on several artificial neural networks. Their results were presented at prestigious international forums. The SZTE also participated in the research of the medical application developed by the Rényi Institue.
Summary of project results:
Science: a total of 79 scientific papers, one patent and one prototype were produced over three years.
Infrastructure: the research infrastructure is in place, typically with high computing capacity servers, which are essential for this research.
Education: during the program, hundreds of university students were introduced to the basics of artificial intelligence by the researchers and educators involved in the consortium.
Innovation: another unexpected result is the emergence of teledermatology. The Covid epidemic made it impossible to take photographs for chronic wound research, but the technology developed – with the help of dermatologists and patients – made it possible to enable remote diagnosis by dermatologists in a very short time. The patient takes a few photos of the area in question, sends them to the doctor, who makes a diagnosis, recommends a therapy and can upload the necessary prescriptions to the cloud – without the patient having to leave the comfort of their home. So far, nearly 20,000 examinations have been carried out in this way.
The next steps
The program has achieved its goal of research into the mathematical foundations of artificial intelligence. This work will be continued on a broader scale by the National Laboratory for Artificial Intelligence (MILAB), also supported by the NKFIH, which will strengthen the synergies between and effectiveness of basic research, applied research and innovation activities.
Hungarian researchers’ new model of ancient global cooling could also help understand the dynamics of climate change today
Hungarian physicists and geologists have modelled the dynamics of one of the most dramatic changes in climate in the history of the Earth, the ancient global cooling that accompanied the glaciation of Antarctica, using a novel combination of laboratory experiments and computer simulations. Their findings, published in Scientific Reports, may shed new light on the cause and effect of an ancient climatic event that affected wildlife 34 million years ago, and thus contribute to understanding the dynamics of the rapid climate change that is taking place today.
The MTA-ELTE Theoretical Physics Research Group and MTA-MTM-ELTE Research Group for Paleontology, both belonging to the Eötvös Loránd Research Network (ELKH), and the Konkoly Thege Miklós Astronomical Institute of the ELKH Research Centre for Astronomy and Earth Sciences (CSFK), the ELKH Institute of Earth Physics and Space Science (FI) and the Department of General and Applied Geology of ELTE investigated the adequacy of the widely accepted textbook explanation of the so-called Eocene-Oligocene transition. The opening of the Drake Passage, the land bridge between South America and the Antarctic Peninsula, played a major role in the history of the event, creating the largest ocean current on the planet, the circumpolar current around Antarctica. The traditional interpretation is that this water cycle isolated the Antarctic from the ocean currents carrying warmth from the tropics, causing the Southern Ocean and Antarctic temperatures to drop until the permanent ice sheet finally appeared.
The researchers studied a radically simplified model of the Southern Ocean in experiments at the Kármán Laboratory of the ELTE Institute of Physics, the only dedicated climate research facility in Central Europe, and at the Brandenburg University of Technology in Cottbus, Germany. The laboratory setup is a cylindrically symmetrical tank rotated on its axis and filled with water, in which the ‘Arctic equator’ temperature difference is provided by heating and cooling the side walls. In the model, a removable bulkhead represented the closed Drake Passage, and high-precision infrared sensors were used to monitor the redistribution of the model ocean temperature after the barrier was removed. Surprisingly, the experiments found that the ‘sea surface’ temperature values did not decrease; instead, they increased in all cases after the opening of the passage, contradicting the textbook argument for the cooling effect of the onset of the circulation.
In order to resolve the discrepancy between the experimental results and the data from historical temperature reconstructions, the team turned to a global computer climate model including the ocean water cycle, which allowed them to virtually ‘close’ the Drake Passage, i.e. to redraw the contours of the continents. The PlaSim (Planet Simulator) software package also allowed equations describing the formation of sea ice to be removed from the model. The results showed that if the simulation allows for the formation of ocean ice, which has a significant reflective and cooling effect, the opening of the Drake Passage will indeed lead to a global temperature decrease, consistent with the Earth’s history. However, when ice is blocked in the program, there is a slight increase in temperature, as in the ice-free laboratory experiment.
The results will contribute to a better understanding of the complex cause and effect of the abrupt climate change of 34 million years ago. The research shows that the traditional argument claiming that the rearrangement of the ocean circulation system alone could have explained the sudden cooling of our planet, which neglects the secondary reflective effect of ice formation, is not valid on its own. Our understanding of the planet’s interconnected oceanic and atmospheric processes in the past can also be used to predict the alarming rate of climate change today.
A new edited volume focusing on post-crisis perspectives of developmental states has been published with contributions from KRTK researchers
“The Post-Crisis Developmental State – Perspectives from the Global Periphery” book edited by Tamás Gerőcs and Judit Ricz, research fellows of the Institute of World Economics, Centre for Economic and Regional Studies (IWE, KRTK) and of the Eötvös Loránd Research Network (ELKH) has been published by Palgrave Macmillan. The focus of the volume is on the role of the developmental state in a situation where a series of major crises affects the (semi-) periphery of the global economy. Several research fellows of the IWE KRTK have contributed to the chapters of the volume.
The authors go beyond the established debate on developmental states in East Asia by highlighting a much broader understanding of development and a very different global economic context. They also further the existing debate by covering new country cases, and at the same time by deepening our perspective on developmental states by looking at unusual sectors such as green industrial policy, education and farming.
Following the financial and economic crisis of 2008-9 and in light of the subsequent prolonged economic recovery, we can observe different types of active state intervention and growing state involvement to reduce the effects of the crisis and to revive economic growth and development throughout the world. Since 2020, this process has been exacerbated by the global coronavirus pandemic in almost all countries of the world, though to a differing extent in each.
Correspondingly, governments must act but under new constraints posed (intensified) by new challenges, which require reconsideration of the repertoire of developmentalist policies and state intervention. Some new economic policy tendencies – such as strengthening economic nationalism and new protectionism – are attempts to respond to these new challenges. The edited volume published in June 2021 aims to map the latest theoretical advances and document recent practical experiences. The studies of the volume try to answer some of the following questions: How can new national development strategies be adapted to newly emerging world economic realities? Which are the most and least successful examples in the world? Is there an existing state-led variety of capitalism which appears to function effectively amidst the new and changing economic world order? How has the repertoire of new – and widely defined – industrial policies changed recently? What role do education and social policies play in the current developmentalist public policy mix? Do development strategies differ in the 21st century in the Global North and South, and specifically in the context of the emerging economies?
The starting point of the volume was the East Asian Miracle of the 20th century, and the classical developmental state paradigm (DSP), which aimed at explaining this outstanding economic catch-up. The classical DSP was the child of its own era, valid only in its original time and space, and embedded in the specific historical and regional context. It is argued therefore that it should not be repeated in a different context, and the emulation of the model is not a relevant option for other countries.
Why is it still relevant to speak about developmental states at the beginning of the third decade in the 21st century? On the one hand, although the classical developmental state model is not applicable today, nor it is functional outside the East Asian region and amidst the new circumstances of the 21st century, it can still provide several historical insights that are relevant even today. To highlight a few lessons, we can mention the central role of “good governance”, conditional state interventions aimed at improving economic performance and international competitiveness, performance-based and meritocratic elite bureaucracies, and good quality education and R&D activities as key components of economic development. The authors of the volume also highlight a further important historical lesson, namely the role of inclusive policies, also known as the stakeholder approach. Inclusive policies are mostly sustainable, as shared economic growth benefiting wider segments of society not only generates multiplier effects for economic growth itself, but also provides legitimacy to the implemented development model and ensures the sustainability of policies. One of the main driving forces of the shared economic growth in East Asia has been the high initial levels and continuous improvement and development of human capital, which also sheds light on the central role of the education system.
On the other hand, it is also worth looking at developmental states today, as at the latest developmentalist state interventions following the 2008-9 global financial crisis – in large emerging economies even earlier –have gained new momentum both in economic policy practice and in related theoretical debates. The developmental state approach is undergoing a new renaissance, yet recent international academic literature is not aiming at reviving the classical DSP, instead, a new concept is under construction, even if a new developmental state paradigm is yet to emerge. Nevertheless, as some of the chapters of the volume illustrate, the reference towards developmental states has emerged in the political rhetoric of several emerging economies ranging from Sub Saharan Africa, Latin America, South Asia to Central and Eastern Europe.
New theoretical trends building upon most recent theoretical advances – in a deductivist manner – differ significantly from the classical paradigm, which was constructed based upon the practical experiences of East Asian economies (via inductive reasoning). The starting point for new developmentalist thinking is the capability approach to development, while reconsidering the main objective, as well as reconfiguring the ways and hows of designing and implementing state interventions. At the same time, these theoretical considerations do not yet or barely spill over into economic policy practices or rhetoric. Strong leaders of numerous developing and emerging economies tend to consider the classical developmental state model – or even some kind of oversimplified interpretation of it – as a role model to emulate, and refer to these simplistic views in their political rhetoric, while highlighting some arbitrarily chosen elements of the classical model.
The studies in the volume point out that no new economic (developmental) success story is emerging to date from recent statist experiments, which would be comparable to the East Asian Miracle, even if it might be too early to make any judgement as we still lack the historical perspective. Nevertheless, the documentation of recent decades’ economic policies and the observation of real-world changes, as well as the comparative analysis of these with more recent theoretical insights, makes it possible to formulate novel conclusions, while it is inevitable that these new insights will be compared with historical experiences and lessons.
Finally, one of the main conclusions of the volume is to emphasize that the aim and character of state interventions is crucial with regard to the potential outcomes. Thus, the historical insight of classical developmental states is still valid: the main goal of state interventions should be to increase competitiveness and export performance via market-friendly measures, to promote technological development, to climbing up in global value chains, and to increase economic embeddedness through intensive forward- and backward linkages. However, success depends on pragmatic implementation: successful measures shall be maintained and extended, while those failing to succeed are phased out and suspended according to the trial and error principle. State interventions shall focus on improved economic performance, economic and social “upgrading” – job creation and qualitative changes. Therefore, a favourable institutional setting and supportive economic policies are of special significance.
The research was supported by the National Research, Development and Innovation Office project entitled “From developmental states to new protectionism: changing repertoire of state interventions to promote development in an unfolding new world order” (FK_124573).
The EU Horizon 2020 Safeguard project, which involves experts from the ELKH Centre for Ecological Research (CER) and the Regional Centre for Information and Development of Knowledge Ltd (RCISD), addresses the decline of wild pollinators and its effects on biodiversity and ecosystem services. The project, which is now underway, will receive €7.5 million in funding from the European Union’s Horizon 2020 research and innovation programme to improve our understanding of the status of and changes to Europe’s wild pollinators, including bees, butterflies, flies and other pollinating insects. The consortium brings together researchers, NGOs, industry and policy experts from 25 institutions in 14 European countries and China.
Wild pollinators are a key component of Europe’s biodiversity, contributing to human well-being in many ways and playing a key role in the functioning of natural ecosystems. They face a number of threats worldwide and in Hungary, including climate change, land use and habitat change. However, the actual extent of the decline and its causal links are still unclear, as is how best to reverse the process.
Reducing multiple pressures on wild pollinators
Under the project, experts will carry out field research to systematically analyse the many factors that threaten wild pollinators and evaluate the effectiveness of different interventions. One of Safeguard’s aims is for participants to develop a range of new methods to support pollinators in stages from the agricultural field to the landscape level – and in agricultural, natural and urban ecosystems.
Mobilising for concerted action
Safeguard will work with key stakeholders to develop an integrated assessment framework, including decision-making protocols, so that its research expertise can better support global, European and national policy makers in their decision-making. The project aims to enhance knowledge of wild pollinators and raise awareness of their value to society, in particular to the general public, industry, business and decision-makers, in order to reverse the decline of wild pollinators through a wide range of coordinated actions across Europe.
“This interdisciplinary project will make a significant contribution to the conservation of wild pollinators in Europe and will strengthen the conservation of domestic pollinators and the international recognition of Hungarian researchers,” said András Báldi, scientific advisor to ÖK and one of the Hungarian participants in the project.
The first part of the ITER central magnet, the world’s first experimental fusion reactor built with the contribution of EK-CER researchers, has arrived
The first part of the world’s first experimental fusion power plant, the gigantic central magnet of ITER, has arrived at the Cadarache center in France. ITER is currently one of the largest international magnetic-fusion research and development projects in the world, and involves experts from the ELKH Centre for Energy Research (EK-CER) Fusion Plasma Physics Laboratory and the Fusion Technology Laboratory, as well as several companies from Hungary.
The aim of the international project involving 35 countries – including Hungary – is to achieve controlled energy production through hydrogen fusion. The 66-ton magnetic part of the Central Solenoid, created by General Atomics in California, arrived by sea from the United States and was then transported to the site by road from the port of Marseille.
The other five parts of the magnet – the assembly of which is an important step for ITER – will be assembled by 2024 at the latest, according to Bernard Bigot, Director of the ITER Organization. When complete, the Central Solenoid will weigh nearly a thousand tons and will be 18 feet high.
“This will be the world’s most powerful magnet, as it will have 13 tesla magnetic fields inside, making it 300,000 times larger than the Earth’s magnetic field,” explained Thierry Schild, Solenoid’s technology manager.
The Hungarian engineers have planned the cabling of the entire interior of ITER in such a way that it can operate without maintenance for 20 years, and they also tested some components in Budapest. EK-CER experts are currently working on the various components of ITER, as well as testing its critical technologies. In the ITER facility, hydrogen gas will be heated to a temperature higher than the temperature in the middle of the Sun, some 150 million degrees Celsius, ‘burning’ it into helium and producing ten times more energy than is used to heat the material.
According to estimates, the experiments will generate the first plasma (ionized hydrogen gas) by December 2025, and full operation is expected by 2035. As a working commercial energy source, fusion energy will not generate electricity before 2050. Proponents of nuclear fusion say it is the energy of the future because it can produce virtually unlimited quantities and does not pollute the environment.
The original budget for the project has already tripled, and currently stands at approximately EUR 20 billion.
An international research group led by CSFK has studied the ancient temperature and precipitation conditions of the last ice age in the southern part of the Carpathian Basin
An international research group led by the staff of the Institute for Geological and Geochemical Research (CSFK FGI) of the ELKH Research Centre for Astronomy and Earth Sciences studied the ancient temperature and precipitation conditions of the last ice age in the southern part of the Carpathian Basin. The research was carried out by examining snail shells preserved in loess sediments and based on the stabilizing isotope compositions of stalactite crusts in nearby limestone recesses. Their latest findings were published in the journal Paleoceanography and Paleoclimatology, published by the American Geophysical Union (AGU).
During the last ice age in the North Atlantic – which was fundamentally a much colder climate compared to today’s conditions – there were significant and rapid fluctuations in climate. “However, we know little about the impact of these rapid climate fluctuations on Europe’s terrestrial climate and ecosystem,” emphasized Gábor Újvári, research leader and CSFK FGI senior researcher. This is mainly due to the fact that accurate primeval temperature estimates for the continent are not readily available from climate archives whose age can be determined with high accuracy.
The research group examined both loess sediments formed from wind-transported dust accumulated during the ice age, and also stalactite crusts located close to it in southern Hungary. The exact time of the formation of these climate archives, which according to the studies ranged from 31,000-26,000 years ago, was determined by radiocarbon and uranium-thorium dating. The formation temperature of the carbonate of the snail shells extracted from the loess sediment was determined by the so-called coupled isotope method (more information on this in a previous piece news on the technique).
The compositions were measured at the Geological Institute of Zurich, ETH, Switzerland, from which the average temperature (TM-OTM-O) characteristic of the mollusc growth period (approximately May to October) could be calculated. According to the analysis, during the cold periods (stadials) that mainly characterize the ice age climate, TM-O ranged from 7 to 14°C, which corresponds to annual average temperatures of between -3 and +4°C (compared to today’s 11°C). During the intervening sudden warm period, TM-O could reach 16–18°C, which assumes annual mean temperatures (9–11°C) similar to that of today. “As a result, during the sudden, brief warm-up periods (interstadials), the temperature rose by about 4-7°C, something that pleasingly was also reflected in the oxygen isotope composition (δ18O) data of the stalactite crust,” says Attila Demény, Director of FGI, head of the stalactite field of research.
The above-mentioned oxygen and coupled isotope analyzes were supplemented by carbon isotope measurements, which made it possible to deduce the aridity (dryness) conditions of the studied area. These show that during the sudden warm periods (interstadials) the summers were drier and warmer, but the year as a whole (and especially the winter period) was wetter, while the stadials were much colder and drier on an annual basis, something which favored dust accumulation. All these rapid climate changes were due to coupled ocean-atmospheric realignments, suggesting that any change in ocean currents in the North Atlantic during the current period of global warming could have a significant impact on Hungary’s climate.
CER researchers demonstrate that grazing can provide effective protection against damage caused by invasive woody species on floodplains
The effect of cattle grazing in poplars on the quantity of invasive woody species, the species richness of the grassland level and the quantity of grazable grasses has been studied in research led by László Demeter, forest ecologist at the ELKH Centre for Ecological Research (CER), and Ábel Péter Molnár, PhD student at the Hungarian University of Agricultural and Life Sciences. The researchers’ paper on the topic has been published in the prestigious Journal of Environmental Management.
Biological invasion is today one of the most significant causes of biodiversity loss and changes to the landscape. The aggressive spread of invasive plant species not only causes a decline in native biodiversity, but also has broader socio-ecological consequences. Many alien woody species settle and spread very efficiently in the floodplains of rivers, often forming dense, almost impenetrable stands. Under the dense, closed stands, most of the native plant species become extinct, no grassland remains, and in the event of flooding, the water only flows very slowly in the overgrown area, significantly increasing the risk of flooding.
In Hungary, but more broadly in most of the floodplains of the Carpathian Basin, these various, rapidly spreading alien plants pose a serious threat to biodiversity. One of the biggest challenges in conservation is to develop effective yet environmentally friendly strategies to prevent and control the spread of plants that make flooding more likely. Though grazing floodplains with domestic animals in combination with other technologies (e.g. mowing, pasture mowing) is one viable alternative, its exact effects are yet to be thoroughly documented. This shortcoming has been met by a study by the authors on the socio-ecological effects of grazing in floodplain noble poplars.
In Hungary, traditional forest grazing systems have almost completely disappeared. As a result, the researchers studied a Serbian section of the Timis River where false indigo bush (Amorpha fruticosa), green ash (Fraxinus pennsylvanica) and boxelder maple (Acer negundo) are found in the floodplain in poplar plantations, while traditional floodplain grazing also takes place.
Researchers have shown that in poplar stands where intensive grazing takes place, the number and abundance of non-native woody species is significantly lower, while grassland levels consisting of more species-rich marsh and meadow species are developing, with more food for cattle. The study also points out that traditional grazing in this landscape results in a ‘win-win-win’ situation, i.e. a triple benefit for society. On the one hand, the traditional floodplain grazing of local communities can be maintained, while, on the other, flooding is easier in shrub-free, grazed floodplain plantations, something which is also beneficial in terms of nature conservation, as grazing helps maintain biodiversity. In addition, grazing of poplars can also be a valuable practice for the forest manager, as it does not damage the stand with proper care, but the harvest is greatly facilitated by the lack of dense alien woody species.
On plantations of newcomer tree species, such as poplar or acacia, the amended Forest Act of 2017 already allows grazing with domestic animals in Hungary, thus removing a significant obstacle to the solution proposed by the researchers.