Speakers

Harnessing Vertically Integrated Projects for Growth in Emerging ICT Fields

Vertically integrated projects bridge the gap between theory and practice, fostering innovation and collaboration across disciplines to solve complex real-world challenges. Simon Fraser University defines experiential learning as “the strategic, active engagement of students in opportunities to learn through doing and reflection on those activities, which empowers them to apply their theoretical knowledge to practical endeavors in a multitude of settings inside and outside the classroom.” This presentation focuses on the real-life classroom teaching experience based on four dimensions of knowledge learning in ICT fields by creating the thrust of Project Based Learning (PBL). In addition, it is established that the projects executed during the study as part of the curriculum provide students with sufficient exposure required to enable them to board on real-world industrial application development projects. To validate this claim, a case study of students demonstrated the impact of experiential learning, whose basic semester-level projects turned out to be innovative, and the impact of this experiential learning becomes remarkable for future careers and our industry, which desperately needs modernization. In addition, this presentation also covers various aspects of the Vertically Integrated Projects (VIP) Program that can provide an opportunity for students to earn academic credit while engaging in authentic and extended research and design projects related to active research areas of ICT and contribute towards developing collaborative projects to get national and international fundings. 

Digital Deception in the AI Era

The rise of artificial intelligence (AI) has brought about significant advancements, but it has also given rise to digital deceptions, including highly sophisticated techniques such as DeepFake technology. This presentation explores various examples of AI-driven deceptions, illustrating how they manipulate audio-visual content to spread misinformation, compromise privacy, and pose ethical challenges. DeepFakes, among other AI-generated forgeries, serve as a focal point, highlighting their potential to undermine trust in digital media. To address these challenges, the latter part of the presentation delves into countermeasures, ranging from fostering AI awareness and ethical use to deploying advanced AI detection tools. These measures aim to equip individuals, organizations, and policymakers with the knowledge and resources necessary to combat AI-driven digital deception effectively, ensuring a safer and more trustworthy digital ecosystem. 

Cryptographic Evolution Meets Internet of Things (IoT): A Quantum Perspective

Kevin Ashton first introduced the concept of the Internet of Things (IoT) in 1999. IoT offers an environment in which people and things (such as smart phones, gadgets and smart machines, etc) can seamlessly interact with each other anywhere and anytime. Kevin anticipated that likewise the Internet, IoT has potential to transform human lifestyle globally. Today, it can be witnessed that IoT has revolutionized human lives globally just by leveraging internet and seamless connectivity of devices, such as smart phones, smart vehicles, smart healthcare equipment, smart gadgets, and industrial machinery, etc. At one hand IoT is offering enormous benefits, while at the other hand it also raises significant security challenges in general and from quantum computing perspective in particular. It is fact that quantum computing brings massive computing power that put at risk the security systems and protocols that are safeguarding IoT eco system. If security challenges that are raised due to quantum computing are not addressed well in timely manner, they may create a chaos in IoT ecosystem. In view of aforementioned facts, this keynote/invited talk investigates the impact of quantum computing on the conventional IoT security, and discusses the vulnerabilities that could be leveraged by quantum computers to break the IoT security. Moreover, this talk also explores/investigates quantum-safe cryptographic solutions to secure IoT ecosystems. 

Deep Learning Techniques for Brain Tumor Classification and Localization 

Brain tumors represent a significant global health challenge due to their high mortality rates across various age groups. Delays in diagnosing BT can lead to fatal outcomes, making timely and accurate diagnosis using magnetic resonance imaging (MRI) essential. Typically, radiologists analyze MRI scans to identify and diagnose tumors. However, manual assessments are prone to errors, time-consuming, and heavily dependent on the expertise of radiologists or neurologists. Existing computer-aided classification models often face limitations in performance and explainability, particularly in clinical neuroscience applications. The lack of interpretability in these "black-box" models causes physicians to question their reliability. The Deep Learning  has the potential to address these issues by advancing neuroscientific research and supporting healthcare tasks.To improve the explainability of deep learning (DL) models and enhance diagnostic support, we propose a novel framework for brain tumor classification and localization. This framework combines existing methodologies with explainability techniques to offer more reliable and interpretable results. The proposed approach employs a pre-trained VGG-19, a custom-trained VGG-19, and EfficientNet architectures. These models are integrated with modified class activation mapping algorithms. Experimental results demonstrate that the pre-trained VGG-19 combined with Grad-CAM outperforms scratch-VGG-19, EfficientNet, and other state-of-the-art DL techniques in terms of classification accuracy, visualization quality, and quantitative evaluations. By improving diagnostic accuracy and offering interpretable visualizations, the proposed framework has the potential to reduce diagnostic uncertainty and enhance the clinical validation of BT classification models.

Machine Learning for Socio-Economic Development: Harnessing Technology for Inclusive Growth

Machine learning (ML) is transforming industries and reshaping the global economic landscape, but its potential to drive socio-economic development remains underexplored. This keynote will examine how ML can act as a catalyst for addressing some of the world's most pressing challenges, from poverty reduction and job creation to health care accessibility and education.

The talk will begin by providing an overview of current ML advancements and their application in low-resource environments. We will explore case studies demonstrating how ML-based solutions have enhanced agricultural productivity, enabled more efficient public services, and empowered small businesses through micro-finance algorithms. Additionally, we will discuss the critical role of data collection and management in underserved regions, emphasizing the need for ethical and inclusive data practices that protect vulnerable populations.

A core part of this presentation will focus on strategies for fostering collaboration between governments, the private sector, and research institutions to leverage ML for sustainable development. We will highlight the importance of building local expertise and adapting models to cultural and infrastructural nuances to ensure that the benefits of technological advancement reach those who need them most.

Cleft Prediction Before Birth Using Deep Neural Networks

In developing countries like Pakistan, cleft surgeries impose significant financial and emotional burdens on families, often accompanied by considerable pain for affected children. This study proposes a machine learning-based approach to predict the likelihood of cleft lip and palate in embryos before birth, offering a potential preventive solution. Data from 1,000 pregnant women were collected across three hospitals in Lahore, Punjab, via a comprehensive questionnaire capturing variables such as gender, parental history, family history of cleft, birth order, number of children, midwife counselling, miscarriage history, parental smoking habits, and physician visits.

Extensive data preprocessing, including cleaning, scaling, and feature selection, was performed to identify key predictive factors. Various machine learning algorithms, including random forest, k-nearest neighbors, decision tree, support vector machine, and a deep neural network-based multilayer perceptron (MLP), were applied to the dataset. Among these, the MLP model demonstrated superior performance, achieving a predictive accuracy of 92.6% on test data. These promising results highlight the potential of using deep learning models to mitigate future cases of cleft lip and palate, enabling early interventions and reducing the burden on families.

AI-Driven Product Innovation: Harnessing LLMs and RAG

The rapid advancement of AI technologies, particularly Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG), is transforming the way innovative products are conceived, designed, and developed. This talk will explore how LLMs and RAG serve as powerful tools to drive innovation and streamline product development. Through practical examples and real-world case studies, we will discuss how the combination of AI-driven generation and knowledge retrieval can effectively address critical challenges such as hallucinations, outdated information, and data silos. Additionally, we will share actionable strategies to reduce LLM costs while maintaining optimal performance.

Brain tumor classification for MR images using transfer learning and fine-tuning 

Accurate and precise brain tumor MR images classification plays important role in clinical diagnosis and decision making for patient treatment. The key challenge in MR images classification is the semantic gap between the low-level visual information captured by the MRI machine and the high-level information perceived by the human evaluator. The traditional machine learning techniques for classification focus only on low-level or high- level features, use some handcrafted features to reduce this gap and require good feature extraction and classification methods. Recent development on deep learning has shown great progress and deep convolution neural networks (CNNs) have succeeded in the images classification task. Deep learning is very powerful for feature representation that can depict low-level and high-level information completely and embed the phase of feature extraction and classification into self-learning but require large training dataset in general. For most of the medical imaging scenario, the training datasets are small, therefore, it is a challenging task to apply the deep learning and train CNN from scratch on the small dataset. Aiming this problem, we use pre-trained deep CNN model and propose a block-wise fine-tuning strategy based on transfer learning. The proposed method is evaluated on T1-weighted contrast-enhanced magnetic resonance images (CE-MRI) benchmark dataset. Our method is moregeneric as it does not use any handcrafted features, requires minimal preprocessing and canachieve average accuracy of 94.82% under five-fold cross-validation. We compare our results not only with the traditional machine learning but also with deep learning methods using CNNs. Experimental results show that our proposed method outperforms state-of-the-art classification on the CE-MRI dataset.

The Power of Data Science in Transforming the Future

In today's rapidly evolving digital landscape, data science has become the cornerstone of innovation in a variety of industries, helping organizations make smarter decisions, streamline processes, and unlock new growth opportunities. Data science is at the forefront of transforming industries and shaping the future. Through the proper utilization of data, organizations can gain important insights that promote innovation, productivity, and development. By integrating machine learning, artificial intelligence, and big data analytics, smarter decision-making, better customer experiences, and process optimization are made possible across the sectors such as manufacturing, healthcare, finance, retail, and many more.

Data science is helping to improve patient outcomes, predict disease outbreaks, and enable precision medicine in the healthcare industry. It aids with risk assessment, trading automation, and fraud detection in the financial industry. Supply chain optimization, demand forecasting, and tailored advice are all advantageous to the retail industry. Data is used by manufacturing organizations to predict maintenance needs and improve operational efficiency.

As technology progresses, data science will play a critical role in solving some of the world’s most pressing problems, ranging from climate change to global health crises. The future of data science is not only about technological advancements; it is also about building a sustainable, equitable, and data-driven world.

21st Century and P-Learning Environment With Emerging Technologies 

The 21st century has seen some emergence new and innovative teaching and learning trends. Among these; Pervasive Learning (P-Learning) is one of the emerging teaching and learning method due to handheld devices’ price reduction, technological support and smartness of smartphone technology. Teachers and learners are now no more restricted by place or time or device and can access digital teaching and learning material whenever and wherever they are. Thus, P-learning has the ability to access teaching and learning material beyond the boundaries of the traditional classroom arrangement. This paper presents the idea of P-learning which is not limited to a single geographic location or mobile or location based technologies; rather, it facilitates teaching and learning from anywhere and at any time with any handheld device means 24*7*12. The purpose of this paper is to propose how this digital teaching and learning paradigm helps physically disabled, geographically scattered, learners through recorded or live audio / video lectures without physically attending or delivering academic classes.