Practical AI Series - IOAI Training

Course Information:
 

Course Title: Practical AI Series - IOAI Training

Year: 2025 - 2026

Target Audience: Form 4 or above

Prerequisite(s): Nil

Learning Hours: 6.0​

​

Course Description:

This series provides a structured introduction to artificial intelligence (AI), covering fundamental concepts, hands-on machine learning, natural language processing (NLP), and computer vision (CV). Through practical coding exercises and real-world examples, students will develop both theoretical understanding and applied skills in AI, preparing them for competitions such as the IOAI and fostering responsible AI literacy.

​

Course Learning Outcomes:

1. Understand core AI concepts - Explain how machine learning, NLP, and computer vision work.

2. Apply AI tools and techniques - Use Python, scikit-learn, and Tensorflow to build and evaluate models.

3. Critically assess AI systems - Evaluate model performance, identify ethical implications, and recognize limitations.

​

Topics:

• L1: Introduction to IOAI & AI in Society

• L2: Machine Learning

• L3: Machine Learning II

• L4: Natural Language Processing and Computer Vision

​

L1: Introduction to IOAI & AI in Society

This session serves as a comprehensive introduction to both the International Olympiad in Artificial Intelligence (IOAI) and the broader, real-world relationship between AI and society. We move beyond technical definitions to explore how AI is built, applied, debated, and governed—emphasizing that AI literacy is a transdisciplinary necessity for the future. Students will explore:

• What is IOAI? – Understand the structure and goals of the STEM-focused competition, focusing on machine learning, NLP, and computer vision. Learn about the Hong Kong Regional Qualifier and access preparation resources for IOAI 2026.

• Narratives & Reality – Analyze dystopian, utopian, and hype-driven narratives, grounding the discussion in AI's current role as a powerful tool that requires informed and critical engagement.

• AI's Building Blocks – Introduce the core components enabling modern AI: Artificial Intelligence (the brain), Cloud Computing (the muscle), and Data (the fuel).

• A Transdisciplinary Tour – Explore real-world AI applications across diverse fields.

• Physical AI – Discover how AI learns, adapts, and acts in the real world.

  • The Perception Stack – Sensors (RGB, LiDAR, Radar, etc.) and Sensor Fusion for holistic environmental awareness.

  • Computer Vision & CNNs – How Convolutional Neural Networks detect patterns through convolution.

  • Vision Language Models (VLMs) – AI that combines vision and language for commonsense reasoning about scenes.

• Societal & Ethical Challenges – Debate critical implications.

• The Ultimate Responsibility – Conclude that AI is an amplifier of human intent, and its future depends on our collective wisdom, ethics, and governance.

Goal: To inspire students to engage with AI both as a competitive discipline (IOAI) and as a societal force, equipping them with a critical, transdisciplinary understanding of AI's capabilities, challenges, and responsibilities.

 

L2: Machine Learning

This session introduces the core principles of machine learning, focusing on how computers learn from data to make predictions and decisions. Students will explore both the theory and practice of building and evaluating ML models, setting the stage for hands-on AI development. Students will explore:

• Key Machine Learning Concepts: Understand the difference between supervised and unsupervised learning, and learn how models are trained and evaluated.

• Common ML Models: Explore Linear Regression, Logistic Regression, and K-Means Clustering.

• Model Evaluation & Improvement: Learn how to assess model performance using confusion matrix and evaluation metrics such as Accuracy, Precision, Recall, and F1-Score, and how to prevent overfitting with techniques like Regularization (L1/L2) and Cross-Validation.

• Hands-On Practice: Apply concepts using Python and Jupyter Notebooks with libraries like Scikit-learn, Pandas, and NumPy.

Goal: To equip students with a practical understanding of machine learning workflows—from data preparation to model evaluation—and prepare them for further exploration in AI and the IOAI competition.

 

L3: Machine Learning II

This session builds directly on the foundations of Session 2, diving deeper into supervised and unsupervised learning techniques while emphasizing hands-on coding practice. Students will move from theory to application, implementing real machine learning models in Python. Students will explore:

• Checkpoint Practice: Complete structured classwork on K-Means Clustering and Logistic Regression, reinforcing learning through practical problem-solving.

Goal: To solidify machine learning knowledge through applied practice, enabling students to confidently build, evaluate, and refine ML models using Python.

 

L4: Natural Language Processing and Computer Vision

This session explores how computers understand human language and visual content, moving beyond numerical data to process text and images. Students will discover the core techniques behind technologies like smart assistants, translators, facial recognition, and self-driving cars. Students will explore:

• Understanding Text with NLP: Learn how computers convert words into meaningful numerical representation (word embedding), which is the foundation of tasks like text classification, language modeling, and machine translation using models like BERT and GPT.

• Seeing with Computer Vision: Discover how digital images are processed as grids of pixels, and how Convolutional Neural Networks (CNNs) use filters to detect patterns, from simple edges to complex objects, enabling applications like object detection and image segmentation.

• Hands-On Implementation: Apply concepts through interactive coding exercises, including building and visualizing a CNN using Jupyter Notebooks and PyTorch/TensorFlow.

Goal: To demystify how AI interprets human language and visual information, providing students with a functional understanding of NLP and CV pipelines and their transformative role in modern AI systems.