Tutorials Speakers of SETIT 2022

Abstract

: One key component of virtually all machine learning methods is optimization of some objective function. Recent methods like deep neural networks require the solution of very large problems, and there is a host of techniques that has been developed for this purpose, both with solid theoretical ground and out of hands-on experience. However, there is more to computational intelligence than just neural networks. Abstract: The Big Tech advocate the use of science to overcome the biological limits of the human body. This new world with technology, where science evolves every day to compensate the deficits of the human body, this new world will most certainly end up creating post-humans: improved men with increased capacities and life expectancy almost infinite. Man is clearly tempted to take power over the environment and over himself. We have ultra-powerful machines, IoT for data collection, storage capacity and of course algorithms, that is to say artificial intelligence. This artificial intelligence can indeed let us believe that man can take power over the environment and over himself. In recent years, machine learning has become an important solution in the healthcare industry. It allows us today to predict a decompensation several days before its occurrence and this is a reality today. Abstract: This tutorial-style presentation will start with a historic overview of Artificial Intelligence (AI), shortly going over the earlier AI waves. The focus will be on the rapid rise of AI in the last decade, narrowing it down to Deep Learning, perceived as an ubiquitous solution for a plethora of applications. This trend was/is stimulated by massive financial support and flourishing on the growth of plenty of custom AI chips. The fast pace rising start-ups in these deceptively esoteric fields will be identified, and their latest results surveyed. Currently, a key ingredient, besides new designs, is extreme ultraviolet lithography EUVL (Extreme_ultraviolet_lithography)— at the heart of fabricating the most advanced few-nanometer integrated circuits (powering cloud, fog, and edge AI & IoT, and most probably quantum computing as well). We will mention some of the technical problems faced, and go over the latest solutions (some landing the German Future Prize in Fall 2020). All of these pinpoints to a monopolistic growth potential revealing extremely stringent financial and technological constraints. Finally, we will conclude by commenting on the forthcoming growth potential of AI hardware in the wider context of rebooting and quantum computing, as seen through the expected demise of Moore’s law. Abstract: In various places of nature, we see certain patterns that repeat themselves after some scaling in size and placement or rotation. These patterns have been studied and modeled by fractal geometries like the Cantor, Koch, Peano, and Sierpinski. There, a certain dimension or angle of the object shape that is being repeated is expressed by a certain mathematical formula that shows the relation between the repeated shapes. On the other hand, some new shapes exhibiting scaling, repetition, and filling have been generated with the help of computer graphics, thus achieving much more complex shapes. The lecture will discuss fractals in nature, former trials to model them, and the mathematical relations that generate new shapes. The lecture will then emphasize the use of fractals in communication engineering. Of special interest is the use of the fractal concept in the design of antennas. It will be shown how the two features of the fractals, scaling and repetition are employed to design wideband antennas and filters. It also aims to propose new fractal concepts that offer flexibility in the design of antennas. A challenge is raised for developing new uses of fractal geometries in the general field of communication engineering. Abstract: We introduce the Layered Ensemble Model which combines Graded Possibilistic Clustering model and ensemble of Artificial Neural Network predictors, obtaining in such a way an accurate forecaster of the traffic flow rates with outlier detection. Experimentation has been carried out on two different data sets. The former consists on real UK motorway data and the latter is obtained from simulated traffic flow of a street network in Genoa (Italy). The proposed model for short-term traffic forecasting provides promising results and given its characteristics of outlier detection, accuracy, and robustness can be fruitful integrated in traffic flow management systems, allowing the local administration to streamline the traffic and reduce traveling time. This will lead to significant energy savings, less pollution, and a better quality of life of the population. traffic forecasting provides promising results and given its characteristics of outlier detection, accuracy, and robustness can be fruitful integrated in traffic flow management systems, allowing the local administration to streamline the traffic and reduce traveling time. This will lead to significant energy savings, less pollution, and a better quality of life of the population. Abstract: Numerous aspects of modern technology have only become realistic as a result of the rapid development of the disciplines Artificial Intelligence and Machine Learning. These two disciplines are expected in the near future Abstract: New generalizations of cellular automata are proposed. Cellular automata had been considered in the restricted regions of space. The cases of external and internal boundaries ware considered. Special rules for cells near boundaries are proposed. Special rules for cells near the boundaries are proposed for gliders. The concepts for modeling logical gates are proposed. For the implementation of logical gates, the propagation of the gliders of cellular automata in bounded domain is proposed. Special rools for collisions of gliders with walls and obstacles are proposed. The realization of logical operations ‘AND’, ‘ÓR’, ‘NOT’, ‘XOR’. Cellular automata on Riemann surfaces are described. Also, it is considered the general formulations and properties of cellular automata with cells which have the strong anticipatory property (introduced by D. Dubois). Multivalued behavior (hyperincursion) of solutions of such CA is describe. It was posed new research problems of computation theory related to presumable multivaluedness of cellular automata with strong anticipation property. Extending of classical automata, Turing machine and algorithms had been proposed. Also, some relation of such cellular automata and quantum mechanics are discussed. Two-Slit computer experiments with cellular automata with strong anticipation are considered. Some applications of cellular automata are described: football; migration on science and high education; epidemic spreading; artificial life. Abstract: Granular media are widely used in many Abstract: Within the field of machine learning, deep learning approaches have resulted in state-of-the-art accuracy in natural language processing. Deep learning techniques hold the promise of emerging technologies. This tutorial is divided into two parts. First, we provide intuitive insights into artificial intelligence, machine learning and focusses mainly on deep learning models and show their applications in natural language processing. We then discuss two case studies on NLP viz BloomNet: A Robust Transformer based model for Bloom’s Learning Outcome Classification and CatBoost: An Ensemble Machine Learning Model for Prediction and Classification of Student Academic Performance. Abstract: Internet Quality of Service (QoS) mechanisms are expected to enable wide spread use of real time services. New standards and new communication architectures allowing guaranteed QoS services are now developed. We will cover the issues of QoS provisioning in heterogeneous networks, Internet access over 5G networks and discusses most emerging technologies in the area of networks and telecommunications such as IoT, SDN, Edge Computing and MEC networking. We will also present routing, security, baseline architectures of the inter-networking protocols and end-to-end traffic management issues. Abstract: Since the beginning of 2020, people, organizations, and governments worldwide have faced several challenges. We ask if we can speak about the circular effects of rapid technological evolution on human behavior. The coronavirus outbreak crisis has disrupted what we all referred to as a ‘normality’ in our daily lives and perturbed the entire world economy. What will the ‘normality’ look like after this challenging time? It is not easy to find answers to this question, and for this reason, we will put into value the importance of the DISPERSAL OF INFORMATION in the relationship between digital technology and human behavior which the coronavirus pandemic has seriously influenced. Finding solutions to this challenge is the purpose of any researchers and practitioners, regardless of their field of interest. All together are seeking solutions to fight against this invisible enemy and re-open the ‘REAL LIFE’ of the people. Abstract: For any territory, knowledge corresponds to information potentially useful to (i) explain and make understandable its internal dynamics as well as its interactions with other adjoining regions in the same or neighboring countries; (ii) manage a region by some local authorities, i.e. by means of some decision-support system, in the spirit of territorial intelligence; (iii) to monitor its daily development through feedbacks and adaptation; (iv) to simulate the future, and design novel projects; and (v) to orient actions for the future. As a consequence, any knowledge base must include the following components (i) geographic objects with their toponyms, characteristics and geometry; (ii) an ontology regrouping types together with topological relations; (iii) a gazetteer regrouping the various names of a place; (iv) some physico-mathem