Educational workshops : ICSPIS 2026 -12th International Conference on Signal Processing and Intelligent Systems

Educational workshops

Workshop Title: Introduction to Deep Learning and Neural Network Implementation with PyTorch

Duration: 4 Hours

By: Amirreza Fateh (PhD Student in Artificial Intelligence, Iran University of Science and Technology)

Workshop Outline:

Overview of Artificial Intelligence, Machine Learning, and Deep Learning concepts
Introduction to the structure and operation of Artificial Neural Networks
Introduction to the PyTorch library and its fundamental concepts
Data preparation and working with Dataset and DataLoader
Implementing and training a simple neural network for image classification
Introduction to Convolutional Neural Networks (CNNs) and their applications
Implementing a CNN model and evaluating its performance
Overview of recent applications of deep learning in computer vision and artificial intelligence

Abstract: Deep Learning is one of the most important branches of Artificial Intelligence and has experienced remarkable progress in recent years. It has been widely applied in fields such as computer vision, natural language processing, healthcare, autonomous vehicles, and intelligent systems. Today, many advanced AI models are developed based on deep neural networks, making familiarity with the concepts and tools of deep learning an essential requirement for students and researchers.

In this workshop, the fundamental concepts of deep learning and neural networks will first be introduced in a simple and intuitive manner. Participants will then become familiar with the PyTorch framework, one of the most widely used tools for developing deep learning models, and will learn the practical steps of data preparation, model design, training, and evaluation. Next, Convolutional Neural Networks (CNNs) and their applications in image classification tasks will be presented, followed by a hands-on implementation example using PyTorch. Finally, an overview of modern deep learning applications and recommended learning paths for further study will be provided.

************************************************************************************************

*****************************************************************************

************************************

Workshop Title: Trustworthy Machine Learning

Duration: 4 Hours

By: Dr. Esmaeel Tahanian (Assistant Professor, Shahrood Univesity of Technology)

Abstract: With the rapid expansion of artificial intelligence applications, achieving high predictive accuracy is no longer sufficient for machine learning models. Today, models are expected not only to deliver strong performance, but also to remain robust against data distribution shifts, provide interpretable and explainable decisions, and maintain resilience against security threats and adversarial attacks. Focusing on the principles of Trustworthy Machine Learning, this workshop introduces the fundamental concepts of generalization, explainability, and security, along with their associated challenges, methodologies, and practical applications.

Generalization

Why Do Models Fail When Encountering Out-of-Distribution (OOD) Data?
Strategies for Enhancing Model Robustness to Out-of-Distribution Data
When Biases Change: Can Models Still Generalize Reliably?

Explainability

Explainability and Interpretability: Why Should Machine Learning Models Be Understandable?
Modern Approaches to Interpreting Machine Learning Model Behavior

Security

Security in Machine Learning Models: Threats and Challenges
Can Machine Learning Models Be Fooled?
Training Data Poisoning: A Threat to Trustworthy Machine Learning
Large Language Models: Opportunities and Security Challenges

************************************************************************************************

*****************************************************************************

**********************************

Workshop Title: From Deep Learning to Foundation Models: The Future of Artificial Intelligence in Neuroimaging with a Focus on Multiple Sclerosis MRI Analysis

Duration: 6 Hours

By: Dr. Mahshid Dehghanpour (PhD in Artificial Intelligence, Researcher and Lecturer, Shahrood University of Technology)

Abstract: Artificial Intelligence is rapidly reshaping medical image analysis and transforming the diagnosis, monitoring, and management of neurological disorders such as Multiple Sclerosis (MS). Recent advances have moved the field beyond conventional deep learning toward foundation models, explainable AI, multimodal learning, and clinically trustworthy intelligent systems. As these technologies continue to evolve, researchers and healthcare professionals need a comprehensive understanding of both current state-of-the-art methods and the future direction of AI in neuroimaging.

In this workshop, participants will explore the evolution of AI for MS MRI analysis, from classical deep learning architectures to the latest medical foundation models and multimodal AI frameworks. Through practical demonstrations, benchmark datasets, and real-world clinical case studies, attendees will gain a clear understanding of modern AI techniques, their clinical applications, current challenges, and emerging research trends. The workshop is designed to provide participants with both the conceptual foundations and practical insights needed to understand, evaluate, and develop next-generation AI solutions for neuroimaging.

Workshop Overview

Evolution of AI in Medical Image Analysis
Deep Learning for MS MRI Analysis
Explainable AI in Clinical Neuroimaging
Foundation Models in Medical Imaging
Benchmark Datasets and Grand Challenges
Multimodal AI for Multiple Sclerosis
The Future of AI in Neuroimaging
Case Study Demonstration

Conference Countdown

3 month(s), and 27 day(s)