Intelligent Information
Technologies for Industry

Rajeev Shorey

Ph.D., Professor, The University of Queensland - IIT Delhi Academy of Research (UQIDAR), IIT Delhi & Adjunct Faculty, IIT Delhi, India

Recent Investigations in Machine Learning and Edge Computing

In this talk, we highlight the research challenges in the intersection of Machine Learning and Edge Computing. More specifically, we look at the Federated Learning paradigm in a faulty edge ecosystem. Federated Learning deviates from the norm of ”send data to model” to ”send model to data''. When used in an edge ecosystem, numerous heterogeneous edge devices collecting data through different means and connected through different network channels get involved in the training process. Failure of edge devices in such an ecosystem due to device fault or network issues are highly likely.

In this talk, we first analyse the impact of the number of edge devices on an FL model and provide a strategy to select an optimal number of devices that would contribute to the model. We observe the impact of data distribution on the number of optimal devices. We then investigate how the edge ecosystem behaves when the selected devices fail and provide a mitigation strategy to ensure a robust Federated Learning technique. Finally, we design a real-world application to highlight the impact of the designed mitigation strategy. The talk will end with a brief discussion of several open research problems in the intersection of Machine Learning and Edge Computing.

Speaker BIO

Dr. Rajeev Shorey is the CEO of the University of Queensland – IIT Delhi Academy of Research (UQIDAR) at IIT Delhi, India. Rajeev also serves as an adjunct faculty in the Computer Science & Engineering department at IIT Delhi. He is an Honorary Professor at the University of Queensland, Australia. Rajeev received his Ph.D. and M.S. (Engg) in Electrical Communication Engineering from the Indian Institute of Science (IISc), Bangalore, India in 1996 and 1991 respectively. He received his B.E degree in Computer Science and Engineering from IISc, Bangalore in 1987. Rajeev career spans several reputed research labs – TCS Research & Innovation, General Motors (GM) India Science Laboratory, IBM India Research Laboratory and SASKEN Technologies. Dr. Shorey served as the first President of NIIT University from 2009 to 2013 before joining the TCS Research Labs in 2014. Rajeev’s work has resulted in more than 75 publications in international journals and conferences and several US patents, all in the area of wireless and wired networks. He has 13 issued US patents and several pending US patents to his credit. Dr. Shorey serves on the editorial board of the IEEE Internet of Things Journal and Springer’s Journal of Wireless Networks. His areas of interest are Wireless Networks including 5G Networks, Telematics, IoT, Industrial IoT, IoT Security and Automotive Networks, including the intersection of AI, Machine Learning and Communication Networks.

For his contributions in the area of Communication Networks, Rajeev was elected a Fellow of the Indian National Academy of Engineering in 2007. He was recognized by ACM as a Distinguished Scientist in December 2014. He was elected a Fellow of the Institution of Engineering and Technology (IET), UK in 2021. Rajeev serves on the Executive Council of ACM India and the Executive Council of IEEE ComSoc, Delhi Chapter. He is the co-founder & steering committee co-chair of two of the highly reputed conferences – COMSNETS (www.comsnets.org) and AIMLSystems (www.aimlsystems.org).

Vladimir Gorodetsky

D.Sc. (Tech.), Professor, JSC Eureca, Russia

Three Knowledge Sources and Three Constituents of Artificial Intelligence Foundation. (Looking Back and Looking Ahead)

The objective of the talk is to justify the various relationships between Data Science as a new one and AI existing and developing in a common interdisciplinary context. We intend to prove that although it is hardly possible to draw a clear boundary between these sciences both they have different research subjects, they have their own specific research methodologies, they solve many different problems, and each of these sciences contributes to strategic IT.

On the other hand, the presentation prove that it is the emergence of Data Science that has greatly contributed to a dramatic increase in the IT industry's interest to the practical use of AI models, methods and technologies. Exactly Data Science has made available, to AI professionals, new, very rich sources of knowledge that is big data. Indeed, this knowledge has become available due to Data Science, due to its methodology, models, methods, technologies and automotive software tools for extraction knowledge hidden in data and access to other kinds of values contained in it. It is well known that AI is unthinkable without knowledge, and Data Science methods, models, algorithms and technologies provide the bulk of the knowledge. Thus, both sciences, Data Science and AI are different sciences, but inseparable, from IT industry point of view, and therein lies the answer to the question posed in the title.

Speaker BIO

Doctor of Engineering (1973), Professor (1990), Honored Scientist of the Russian Federation (1992), leading researcher of JSC EURICA. Received Master degree in mechanics from Leningrad Air Force Engineering Academy (1960) and Master degree in mathematics from Leningrad State University, Mathematics and Mechanics Faculty (1970). Defended PhD (1967) and doctoral (1973) dissertations on optimal control problems. From 1967 to 1988 he worked at the Military Engineering Academy after Mozhaisky in Leningrad in research and teaching positions. From 1988 till 2018 he worked at the St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences in the positions of Head of Laboratory and Chief Scientist of the Laboratory of Intelligent Systems. He has published more than 250 papers (over 100 in English), 9 monographs and textbooks. He served as co-editor of 12 issues of Lecture Notes in Artificial Intelligence and Lecture Notes in Computer Science series (2001-2015), guest editor of special issue of IEEE Intelligent Systems (Q1) on "Agent and data Mining Interaction" (2009), etc. Member of Russian and European Artificial Intelligence Associations, IEEE, IEEE Computer Science, International Foundation for Autonomous Agents and Multi-agent Systems (IFAAMAS), International Society of Information Fusion (ISIF). Associate Editor of the international journal Data Science and Advanced Analytics (Springer, Q1) and member of editorial board of the Russian journal Design Ontologies. He is the recipient of the D.A. Pospelov Award of Russian Association for Artificial Intelligence (2022).

Scientific data (as of 01.01.2023): RSCI h-index=24, WoS h-index=7, Scopus h-index=12.

Areas of Research Interests: Optimal control theory, applied statistics, intelligent planning and scheduling, artificial intelligence and decision making, multi-agent systems, group control, self-organizing agent networks, software tools, data mining and machine learning, efficient and robust big data processing algorithms, semantic computing, distributed and p2p learning, computer and information security, dealing with uncertainty, recommendation systems, transportation and manufacturing logistics

Web page: https://ru.wikipedia.org/wiki/Городецкий Владимир Иванович (in Russian)

Alexey Karpov

D.Sc. (Tech.), Professor, ITMO University, St. Petersburg Federal Research Center of RAS, Russia

Intelligent interfaces and systems for human-computer interaction

In this talk, I will briefly analyse a state-of-the-art in the area of intelligent interfaces and systems aimed for organization of natural human-computer interaction. I will provide major world-science trends in the direction of designing intelligent systems based on digital processing of multimodal information (audio, video, and text) coming both from users to systems (intelligent analysis) and from systems to users (intelligent synthesis). In detail I will consider some samples of intelligent interfaces and systems for natural human-computer interaction that have been recently developed in the Speech and Multimodal Interfaces Laboratory of SPC RAS and aimed for complex multimodal analysis of human behaviour including recognition of human’s affective psycho-emotional states and personality traits with regard to actual tasks in industry.

Speaker BIO

Dr.Sc.(Tech.), Professor, Head of the Speech and Multimodal Interfaces Laboratory of St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences at the St. Petersburg Federal Research Center of the Russian Academy of Sciences, as well as a part-time Full professor at the Faculty of Information Technologies and Programming of ITMO University. He has over 20 years of scientific research experience in the areas of artificial intelligence and intelligent human-computer interfaces and system. He has published over 350 scientific papers in leading journals and proceedings, including 4 monographs. His H-index is 19 according to Scopus and 30 according to Google Scholar. He is a general co-chair of a series of top international conferences “Speech and Computer” SPECOM. He also supervises a Leading scientific school of Russia, as well as leads research projects from RSF, RFBR and the Ministry of Science and Higher Education of Russia.

Lev Utkin

D.Sc. (Tech.), Professor, Peter the Great St. Petersburg Polytechnic University, Russia

Attention mechanism in machine learning models by tabular training data

The attention mechanism is an effective method for improving the performance of neural networks. It has been successfully applied to many important tasks, including the natural language processing, computer vision, etc. In spite of the success, attention is a component of neural architectures and, therefore, the corresponding models meet difficulties of neural networks, including, overfitting, many tuning parameters, requirements of a large amount of data, the black-box nature, expensive computations. Moreover, tabular learning data may be also a problem encountered with neural networks. Therefore, new approaches to incorporate the attention mechanism into such models as random forests, the gradient boosting machine are proposed. The main idea behind the approaches is to apply the Nadaraya-Watson kernel regression model, which leads to outperforming results. Extensions of the attention mechanism, including self-attention and multi-head attention, are also considered. New models can be regarded as alternatives to the conventional attention mechanism when dealing with tabular training data.

Speaker BIO

Professor of the Higher School of Artificial Intelligence in Peter the Great St. Petersburg Polytechnic University, Saint-Petersburg, Russia. Professor, DSc. Head of the Research Laboratory of Neural Network Technologies and Artificial Intelligence in the same university. In 1986 he graduated from St. Petersburg State Electrotechnical University (former Leningrad Electrotechnical Institute). He holds a Ph.D. in Information Processing and Control Systems (1989) from the same university and a D.Sc. in Mathematical Modelling (2001) from St. Petersburg State Institute of Technology, Russia. Awarded an Alexander von Humboldt Foundation Fellowship (2001-2003). Member of the Society for Imprecise Probability Theory and Applications (SIPTA) and the International Society on Multiple Criteria Decision Making (ISMCDM). Author of more than 300 scientific publications, including AI journals: Neurocomputing, Neural Networks, Knowledge-Based Systems, Applied Soft Computing, AI in Medicine, etc. Research interests are focused on machine learning, imprecise probability theory, decision making.

Petr Skobelev

D.Sc. (Tech.), Professor, Samara Federal Center of Russian Academy of Science & Samara State Technical University, Russia

Emergent Intelligence: from the concept - to applications

The paper gives a new definition of the concept of "Emergent Intelligence (EI)", discusses the principles of designing and approaches to the development of EI systems, provide a brief description of the developed models, methods and algorithms for collective decision-making in EI, discusses the functions and architectures of the developed EI systems and presents the results of their application.

Speaker BIO

Professor of Samara State Technical University, Chief Scientist of Samara Federal Center of Russian Academy of Science, Founder and Chairman of Directors Board of “Knowledge Genesis” Group (Skolkovo-Moscow-Samara-Helsinki), Member of International Advisory Board of Czech Institute of Informatics, Robotics and Cybernetics (Czech Technical University), Co-Chairman of International Science and Education Center “Future of Engineering”.

Prof. Petr Skobelev develops Artificial intelligence (AI) for solving complex problems of adaptive resource management for airspace, manufacturing, railways, transport, supply chains, etc. The results of this work help enterprises to make a step towards Industry 5.0 / Society 5.0, based on digitalization of knowledge, formation of digital eco-systems of AI colonies and digital transformation of enterprises to talent-centric networks. “Knowledge Genesis” Group is one of the TOP-10 most innovative and fast growing companies in Russia (by rating of RVC any and PricewatehouseCoopers).

INDUSTRIAL APPLICATIONS: scheduling tankers for Tankers International (UK), taxi for Addison Lee (UK), trucks for GIST (UK), etc.; scheduling cargo for the International Space Station for Rocket & Space Corporation “Energia” (Russia); supply chain scheduling for Lego (USA), Coca-Cola (Germany), SUEK and GasPromNeft (Russia); inter-city long-distance Full-Truck-Load (FTL) cargo scheduling and in-city real-time Less-Than-Truck Load (LTL) scheduling for Coca Cola (Russia) and a number of other transportation companies; agile workshop scheduling for “Axion”, “AviaAgregat” and “Kuznetsov” (air jet plants in Russia) as well as for Airbus and Iacobucci factories (under EU FP7 Program “Factory of Future”); operational scheduling of high-speed Sapsan and 700 other trains for Moscow-Saint-Petersburg, capacity planning for Baikal-Amur Mainline in Siberia and adaptive re-scheduling of Moscow circle line trains; adaptive management of a swarm of satellites for Earth remote sensing (video of swarm of satellites: https://www.youtube.com/watch?v=r7vKK9XnTCE&t=94s\), etc.

DISTINCTIONS: member of Russian National Association of Artificial Intelligence; member of IEEE Association “Industrial Agents”; awarded by National Prizes for Best Company, Best Products, Best Innovations, Export Activities, etc.; chief Editor of “Design Ontology” – the first SemanticWeb journal in Russia (second in EU); member of PC and Invited Speaker of EU MAS conferences: HoloMAS, PAAMS, ICAART, etc.; awarded as a “Best scientist” of Samara city and Samara region; member of Samara Region Governor Council on science, innovation and technology.

More than 15 R&D projects are implemented for RFBR and Ministry of Science and Education under Prof. Petr Skobelev guidance and 5 PhD Students got their degree. He is author more that 230 publications (about 100 – in Scopus), 4 books, published in UK and Germany, 7 textbooks, 25 software products registered, 3 inventions are made in UK and USA.

Summary of R&D results could be found in a book by G. Rzevski and P. Skobelev “Managing complexity” published by WIT Press in UK and USA (is available on Amazone web-site).

Prof. Petr Skobelev h-index is 26 in Google Scholar.