Duration: 2 hours   –   Date and time: This tutorial has been cancelled

Abstract: A brain-computer interface (BCI) enables a user to communicate directly with an external device, e.g., a computer, using brain signals. It can be used to research, map, assist, augment, or repair human cognitive or sensory-motor functions. A closed-loop BCI system performs signal acquisition, temporal filtering, spatial filtering, feature engineering and classification, before sending out the control signal to an external device. Transfer learning (TL) has been widely used in EEG-based BCIs to reduce the calibration effort for a new subject, greatly increasing their utility. This tutorial describes how TL can be considered in as many components of a BCI system as possible, and introduces a complete TL pipeline for EEG-based BCIs. Deep learning approaches are also fully evaluated with classic and state-of-the-art domain adaptation approaches with end-to-end neural networks for both offline and online BCI applications. Examples on multiple datasets demonstrate the advantages of considering TL in multiple components of EEG-based BCIs. Especially, integrating data alignment and sophisticated TL approaches can significantly improve the classification performance, and hence greatly reduces the calibration effort.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: Today computing devices like laptops, mobile phones, smart devices, etc., have penetrated very deep into our modern society and have become an integral part of our daily lives. Currently, more than half of the world’s population uses computers/mobile devices for their professional/personal needs. However, these computing devices are targeted by malware designers encouraged by profits/gains associated with the attack. According to a recent report, monetary losses due to cybercrime are expected to reach 10 trillion dollars annually by 2025. The primary role in providing defense against malware attacks is designed and developed by the anti-malware community (researchers and the anti-virus industry). Traditionally anti-viruses are based on the signature, heuristic, and behavior based detection engines. However, these engines are unable to detect next-generation polymorphic and metamorphic malware. Thus researchers have started developing malware detection engines based on machine learning to complement the existing anti-virus engines. However, there are many open research challenges in these models like adversarial robustness, explainability, fairness, etc., which we are going to discuss in detail during the tutorial.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: Combinatorial problems refer to those applications where we either look for the existence of a consistent scenario satisfying a set of constraints (decision problem), or for one or more good/best solutions meeting a set of requirements while optimizing some objectives (optimization problem). These latter objectives include user’s preferences that reflect desires and choices that need to be satisfied as much as possible. Moreover, constraints and objectives (in the case of an optimization problem) often come with uncertainty due to lack of knowledge, missing information, or variability caused by events, which are under nature’s control. Finally, in some applications such as timetabling, urban planning and robot motion planning, these constraints and objectives can be temporal, spatial or both. In this latter case, we are dealing with entities occupying a given position in time and space.  Because of the importance of these problems in so many fields, a wide variety of techniques and programming languages from artificial intelligence, computational logic, operations research, and discrete mathematics, are being developed to tackle problems of this kind. While these tools have provided very promising results at both the representation and the reasoning levels, they are still impractical to dealing with many real-world applications. Using the Constraint Satisfaction Problem (CSP) formalism, we will explore several exact and approximate solving techniques to address the challenges and limitations listed above.

Duration: 2 hours   –   Date and time: This tutorial has been cancelled

Abstract: With the development of quantum technology, quantum computation has shown powerful computation capability. An emerging area of quantum machine learning has attracted wide interest. In quantum machine learning, the unique characteristics of quantum mechanics is utilized to provide potential advantages for machine learning tasks. In this tutorial, we first present a brief introduction to the basic principles of quantum computation. Then we introduce several quantum machine learning algorithms including quantum neural networks, quantum reinforcement learning and quantum-inspired learning algorithms.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: Humans are social beings and people cannot live alone. Computational social simulation is a way to reproduce a real-world society and study the behaviour of people in that society using computer-based systems. Computational social simulation is a long-term, cutting-edge topic in the interdisciplinary field where information technology, computer science, social science, and sociology overlap. Role-Based Collaboration (RBC) has been proposed as a computational approach to facilitating collaboration. It utilizes roles as underlying mechanisms to support collaboration by taking advantage of roles. It is divided into several phases: role negotiation, role assignment, role execution, and role transfer. RBC and its related components are an abstract model, which is a perfect mapping for social activities, because Social and economic systems are typical collaboration systems. The Environments – Classes, Agents, Roles, Groups, and Objects (E-CARGO) model, which has been developed into a general model for complex systems have a good match for the requirements of computational social simulations. In this talk, we establish the fundamental requirements for social simulation and demonstrate that RBC, E-CARGO, and the subsequent Group Role Assignment (GRA) optimization model are highly qualified to meet these requirements. Based on E-CARGO and GRA, we present a new approach to social simulation with E-CARGO related components, models, and algorithms. This tutorial also illustrates several interesting case studies of computational social simulations.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: The subject of this tutorial links Natural Language Processing, Interaction and linguistics analysis and knowledge representation. Researchers and PhD students that want to discover social media, e-mails and document analysis and cognitive sharing are interesting on this area. In fact, AI research use generally NLP and TextMining and statistics, or Knowledge engineering and semantic representation. But they are aware, Interactions analysis that uses Prama-linguistics techniques belong to social and human science. In this tutorial, we present new techniques based on linguistics and pragma-linguistic that help to identify not only semantic relations but also actors’ intentions. We show also how to combine these techniques with knowledge engineering in order to restitute the meaning of interactions. Examples on analyzing websites, e-mails and discussion forums will be presented.  Finally, Tutorial attendees will apply some of this analysis techniques in given exercises.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: The ultimate goal of multi-objective optimization is to help a decision maker (DM) identify solution(s) of interest (SOI) achieving satisfactory trade-offs among multiple conflicting criteria. This can be realized by leveraging DM’s preference information in evolutionary multiobjective optimization (EMO). No consensus has been reached on the effectiveness brought by incorporating preference in EMO (either a priori or interactively) versus a posteriori decision making after a complete run of an EMO algorithm. In this tutorial, I will present a series of experimental results show that preference incorporation in EMO does not always lead to a desirable approximation of SOI if the DM’s preference information is not well utilized, nor does the DM elicit invalid preference information, which is not uncommon when encountering a black-box system. To a certain extent, this issue can be remedied through an interactive preference elicitation. Last but not the least, we find that a preference-based EMO (PBEMO) algorithm is able to be generalized to approximate the whole PF given an appropriate setup of preference information.

Duration: 2 hours   –   Date and time: This tutorial has been cancelled

Abstract: Smart contract introduced by Ethereum opened the door for the development of various decentralized applications and even decentralized autonomous organization. This tutorial consists of four parts: (1) introduction to blockchain; (2) programming smart contract with Solidity; (3) advanced topics in smart contract development; and (4) vulnerabilities and attacks on smart contracts. Part I will cover the history, the fundamental design principles, and the nuts and bolts of the blockchain technology, including keys, addresses, transactions, blocks, cryptographic primitives, proof-of-work, proof-of-state, and double-spending attacks. Part II will cover the basic of the Solidity programming language and the development environment for smart contracts. Part III will cover several advanced topics in smart contract development, including the application binary interface, the design patterns for smart contracts, and how to save on gas consumption in smart contract. Part IV will cover the incidents that have happened to poorly designed smart contracts in Ethereum, the attack vectors, and the best practices in securing smart contracts.

Duration: 4 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract:

One defining characteristic of twenty-first century engineering challenges is the breadth of their scope.  Each is so large and complex in its own right that each might seem entirely intractable.  Furthermore, each goal might appear so different from the next that one might naturally conclude that the skills needed to solve one challenge are entirely distinct from those of another.  Consequently, our engineering education system would have to turn “on a dime”, orient itself towards each of these 14 challenges, and ask our engineering students to commit themselves to one of these challenges; never to change direction again.  And in the event that we are successful on such a course, the engineering education system would have to pivot again years later to address the newly cropped-up grand challenges.  Quite fortunately, the developing consensus across a number of STEM fields is that each of these goals is characterized by an “engineering system” that is analyzed and re-synthesized using a meta-problem-solving skill set.  In essence, our formidable challenge is one of convergence towards abstract and consistent methodological foundations for engineering systems, in general.  Two fields in particular have attempted to traverse this convergence challenge:  model-based systems engineering (MBSE) and network science.   MBSE has developed as a practical and interdisciplinary engineering discipline that enables the successful realization of complex systems from concept, through design, to full implementation.  Despite its many accomplishments, MBSE’s reliance on graphical modeling language ultimately requires additional mathematical tools to gain quantitative insight.   In contrast, the network science community (NSC) has developed to quantitatively analyze networks that appear in a wide variety of engineering systems.  And yet, despite its methodological developments in multi-layer networks, the NSC has often been unable to address the explicit heterogeneity often encountered in engineering systems. This tutorial serves to introduce the audience to hetero-functional graph theory drawing on several recent publications and a new consolidating textbook entitled:  Hetero-functional Graph Theory for Interdependent Smart City Infrastructures by W.C. Schoonenberg, I.S. Khayal, and A.M. Farid.  It demonstrates that HFGT can be applied extensibly to an arbitrary number of arbitrarily connected topologies of “convergent” engineering systems.  To the MBSE community, we hope that HFGT will be accepted as a quantification of many of the structural concepts found in MBSE languages like SysML.  To the NSC, we hope to present a new view as to how to construct graphs with fundamentally different meaning and insight.  Finally, it is our hope that HFGT serves to overcome many of the theoretical and modeling limitations that have hindered our ability to systematically synthesize, analyze, and re-synthesize the structure and function of convergent engineering systems.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: The Internet of Things (IoT) paradigm enables various smart objects to be connected, this allows us to interact with our environment in an intelligent way. It is believed that low power and ultra-low power sensors would outnumber any other IoT devices by 2030. To achieve the full potential of IoT applications, AI techniques are required to analyze the sensor data on the edge for real-time analytics, reduced latency, and less privacy concern. Low power consumption edge AI also creates significant opportunities for green AI and sustainable IoT research, which aligns with the conference’s theme of “Improving the quality of life”.  In this tutorial, we will first provide a comprehensive overview of low power sensors and compare various IoT communication protocols. We will walk through the detailed end-to-end data integration steps from sensors to the loud data platform using real life award-winning smart cities examples. Since low power sensors are constrained by power and processing resources, integration with computationally intensive Machine Learning (ML) for intelligent processing and decision making becomes a unique challenge. This tutorial will review various methods for applying ML and deep learning to low power sensor solutions. Different hardware and software options will be discussed including bio-inspired chipsets, traditional centralized learning, federated ML, pruning and TinyML for edge computing. We will demonstrate the latest design of our acoustic sensor with edge ML capability for real time sound classification. Development trend and future research opportunities for edge AI and IoT will also be presented.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: A driving force for the realization of a sustainable energy supply is the integration of renewable energy resources. Due to their stochastic generation behaviour, energy utilities are confronted with a more complex operation of the underlying power grids. Additionally, due to technological developments, controllable loads, integration with other energy sources, changing regulatory rules, and market liberalization, the system’s operation needs adaptation. Proper operational concepts and intelligent automation provide the basis to turn the existing power system into an intelligent entity, a smart grid. While reaping the benefits that come along with those intelligent behaviours, it is expected that system-level developments and testing will play a significantly larger role in realizing future solutions and technologies. Proper validation approaches, concepts, and tools are partly missing until now. This tutorial aims to tackle the above-mentioned requirements by introducing validation methods and tools for validating smart grids and energy systems.

Duration: 2 hours   –   Date and time: Oct 1, 2023   –   Location: TBD

Abstract: Many robotic systems involve highly nonlinear, complex, and large-scale decision-making problems in safety-critical situations. Stability and safety are often research problems of control theory, while conventional black-box AI approaches lack much-needed robustness, scalability, and interpretability, which are indispensable to designing control and autonomy engines for safety-critical aerospace and robotic systems. However, the existing performance guarantees of black-box AI approaches are inadequate and often times lead to safety compromises. This tutorial session gives a tutorial overview of machine learning control systems with safety and stability guarantees. We will present some recent results using contraction-based incremental stability tools for deriving formal robustness and stability guarantees of various learning-based and data-driven control problems, with some illustrative examples including learning-to-fly control with adaptive meta learning, learning-based swarm control and planning synthesis, and optimal motion planning with stochastic nonlinear dynamics and chance constraints. We will also present recent results on neural-network-based contraction metrics (NCMs) as a stability certificate for safe motion planning and control.