Workshop Title: On EEG Signal Processing Using the EEGLAB Toolbox

Duration: 2 Hours

By: Dr. Zeynab Mohammadpoory (Assistant Professor, Shahrood Univesity of Technology)

Workshop Outline:

• Introduction to the EEGLAB Toolbox and Data Import
• Preprocessing and Artifact Removal
• Time and Frequency Domain Processing
• Event-Related Potentials (ERP) Analysis
• Creating a Study in EEGLAB and Group Analysis

Abstract:  EEG (electroencephalography) signals are crucial for diagnosing and treating various neurological and psychological disorders, as well as for researching brain function during neuro-cognitive tasks. Additionally, these signals are used for controlling smart orthotics and prosthetics for individuals with disabilities, neuro-marketing, and designing therapeutic and recreational games. Several software tools are available for EEG signal analysis, among which the EEGLAB toolbox in MATLAB is notable for its capability to perform a wide range of EEG signal processing tasks. The EEGLAB toolbox features a user-friendly graphical interface, allowing researchers to process and analyze signals without needing to write code.

This workshop will cover the fundamental concepts and terminologies in EEG signal processing, followed by practical session on common preprocessing and processing techniques for EEG signals.

Workshop Title: Detection of Electrical and Mechanical Faults in Electric Machines Using Signal Processing of Electrical and Mechanical Signals in Transient and Steady States

Workshop Duration: 2 hours

By: Dr. Mohammad Hossein Tabar Marzbali (Associate Professor, Department of Electrical Power Engineering, Shahrood University of Technology)

     Eng. Reza Bazghandi (MSc in Electrical Engineering, Shahrood University of Technology)

Workshop Outline:

 Part One:

• Examination of various electrical and mechanical faults in electric machines.
• Identification of the impact of electrical and mechanical faults on electrical and mechanical signals.
• Investigation of electrical and mechanical faults in the time, frequency, and time-frequency domains.
• Pre-processing of electrical signals to improve fault detection methods.

Part Two:

• Practical implementation of fault in the laboratory.
• Implementation of frequency-based signal processing methods and time-frequency-based signal processing methods on practical electrical signals in both steady-state and transient conditions.

Abstract: Induction machines, due to their low cost and high reliability, have been widely used in the industry. These machines are a key component in modern production industries, with rated powers ranging from less than one horsepower to several megawatts. The strength and low maintenance requirement of induction machines make them ideally reliable for use in many industrial processes. However, electrical and mechanical faults in these machines can incur irrecoverable economic losses. Monitoring the condition of these machines can lead to the early detection of faults, which is essential to minimize such losses.

Based on recent research, electrical characteristics such as stator and rotor electrical currents (in wound rotor induction machines), instantaneous voltage and power, and mechanical characteristics such as speed, torque, vibrations, and magnetic flux are used for detecting various types of electrical and mechanical faults in induction machines, such as rotor imbalance, misalignment, inter-turn short circuit, bearing faults, high resistance connection faults, and gearbox faults. Signals obtained from various electrical and mechanical sensors can be processed using a wide range of signal processing techniques to extract specific fault features. Many of these techniques are based on frequency domain analysis or direct signal processing. After extracting specific fault features, these indicators need to be further processed to not only detect the presence of a fault but also determine its severity. This workshop will focus on teaching the basic principles of condition monitoring along with the implementation of signal processing methods. In this regard, signal processing methods will be implemented on practical data obtained from various faults created on machines using MATLAB software.

Workshop Title: Psychological Profile Extraction from Cyber Space

Duration: 2 Hours

(By: Dr. Maryam Saidi (Assistant Professor, Shahrood Univesity of Technology

Workshop Outline:

• Introducing cyberspace and digital footprints
• Various data available in the cyberspace and the their importance
• Introduction of emotion and mood
• Quantifying emotion from cyber data
• Introducing and quantifying personality
• Extracting people's personality based on cyber data
• Introduction of feeling, sentiment, opinion and quantifying methods of opinion mining
• Introducing some examples of tools for emotional analysis, personality recognition based on data published in cyberspace and working with them.

Abstract: Due to the huge amount of data and information in the cyber space, various organizations and centers are increasingly extracting information from these data for their strategic plans. Extracting features and indicators in the field of cyberpsychology is useful information for many purposes. For example, In order to increase the effectiveness of the content and motivate users to share the desired content, it is better to explore people's thoughts and opinions directly. In this course, psychological indicators including personality, emotion, sentiment, opinion, etc., which can be extracted from the cyber space are introduced. These psychological concepts are explained and the method of quantifying these qualities is introduced. And at the end, we introduce the examples of tools and applications that are provided to extract this information. 

Workshop Title: From Technology to Application: The Role of Graph Neural Networks in Banking Data Analysis

Duration: 2 Hours

By: Zahra Nourollah (Ph.D. Candidate in Computer Engineering - Artificial Intelligence, Shahrood University of Technology)

Workshop Outline:

• Introduction to Graph Neural Networks (GNNs)
• Fundamentals of Technology and Structure of Graph Neural Networks
• Comparison of Graph-Based and Traditional Methods in Data Analysis
• Case Studies
• Challenges and Opportunities in Implementing GNNs in Banking
• Future Developments in Graph Neural Networks and Banking

Graph Neural Networks (GNNs), as one of the most advanced tools in artificial intelligence, enable the analysis of complex data and relationships within graph structures. By modeling connections between various entities—such as customers, transactions, or social networks—this technology can uncover hidden patterns and behaviors. In industries, particularly in domains like banking, GNNs have broad applications. These models play a crucial role in detecting suspicious activities (e.g., money laundering), predicting customer credit scores, and reducing credit risks.

This workshop begins by introducing the foundational concepts of Graph Neural Networks (GNNs) and the importance of modeling data in graph structures. It then explores diverse GNN architectures and popular tools for their implementation. In practical sessions, the applications of GNNs in detecting suspicious money-laundering activities and predicting customer credit scores are explained. Through case studies, the workshop highlights the effectiveness of these methods in reducing credit risks and improving analysis efficiency. Combining technological and business perspectives, this workshop provides a clear pathway for leveraging Graph Neural Networks to enhance banking processes and drive innovation in data analysis