SC310005_ArtificialIntelligence_2025s1

🎓 SC310005: Artificial Intelligence 2025 — Khon Kaen University

🧠 Bachelor-level course on modern AI: Vision, LLMs, and practical AI for real-world impact.

👨‍🏫 Lecturer

Teerapong Panboonyuen or P’Kao

Senior AI Research Scientist, MARSAIL and PostDoc Fellow, Chula

📧 teerapong.pa@chula.ac.th 🌐 GitHub Profile

🧭 Course Overview

This course introduces modern Artificial Intelligence (AI) concepts using real-world datasets and industry-grade tools. Students will explore classical ML, deep learning, computer vision, large language models (LLMs), and AI agents — building practical projects along the way.

✨ Final outcome: Build a complete AI solution — from data to demo — using Python, Colab, and modern AI APIs.

🗓️ Weekly Schedule (16 Weeks)

🔁 Python Recap (for students needing a refresher)

📚 Optional but recommended: Review core Python, NumPy, and Pandas before diving into lab work.

🔍 Covers: Python syntax, NumPy arrays, Pandas DataFrames, data exploration, and basic plotting — all essential for AI programming.

📅 Week 1: Introduction to AI + Python Refresher (NumPy & Pandas)

🎯 Goal: Understand the course structure and refresh Python basics for data manipulation.

🛠️ Lab Activity: Fictional Characters Dataset Analysis

🔗 Files:

💡 Task: Use pandas to explore the Fictional Characters dataset. Clean the data, visualize features, and attempt to build a simple rule-based classifier.

📅 Week 2: Feature Engineering for AI

🎯 Goal: Learn how to transform raw data into meaningful features to improve model performance.

🛠️ Lab Activity: Feature Creation & Selection with Fictional Characters Dataset

🔗 Files:

💡 Task: Engineer new features such as interaction terms, binary flags, or normalized attributes. Evaluate feature importance and prepare the dataset for supervised learning.

📅 Week 3: Supervised Learning with Machine Learning

🎯 Goal: Understand the fundamentals of supervised learning and build classification models using labeled data.

🛠️ Lab Activity: Cancer Classification Model Development

🔗 Files:

💡 Task: Train a supervised machine learning model to classify cancer types based on medical data. Evaluate model accuracy and performance metrics using real-world features.

📅 Week 4: Deep Learning

🎯 Goal: Learn the fundamentals of deep learning and apply Vision Transformers for facial classification using the Thai Prime Minister’s dataset.

🛠️ Lab Activity: Thai Prime Minister Face Classification using Vision Transformer

🔗 Files:

💡 Task: Train a Vision Transformer model to classify facial images of Thai Prime Ministers. Evaluate the model’s accuracy and assess the effectiveness of Vision Transformers in image classification tasks.

📅 Week 5: Recommender System

🎯 Goal: Understand the principles behind recommendation systems and apply them to real-world retail data to suggest products in a Book Store setting.

🛠️ Lab Activity: Building a Book Store Recommender System

🔗 Files:

💡 Task: Apply recommendation system techniques such as collaborative filtering and content-based filtering to suggest books to users in a virtual bookstore environment. The provided dataset contains online retail transactions which must be adapted and used to simulate recommendation logic.

📅 Week 6: Thai Stock Time Series Forecasting

🎯 Goal: Learn and apply time series forecasting techniques on real Thai stock data to predict daily closing prices with the aim to minimize forecasting errors.

🛠️ Lab Activity: Forecasting 5 Thai Stocks

🔗 Files:

💡 Task: Use any forecasting method (moving average, machine learning, deep learning) to model and predict stock prices of these 5 Thai stocks: AOT, BDMS, BAY, ESSO, HMPRO. Evaluate your models with RMSE and MAE and visualize the forecasts versus actual prices.

📅 Week 7: How to Prepare an AI Poster Presentation

🎯 Goal: Each group will consult with the instructor via Zoom to receive feedback and guidance on their poster presentation progress.

🛠️ Activity: Group-by-Group Poster Consultation (Zoom Only)

🔗 Files:

🕐 Note: Please join Zoom at your group’s scheduled time. Be ready to share your current poster draft and any questions you may have.

💡 Reminder: Please be punctual and respectful of others’ consultation time. This session is your chance to refine your project before the final presentation.

📅 Week 8-9: Poster Showcase, Oral Presentation & Midterm Exam

🎯 Goal: Demonstrate your creativity, teamwork, and understanding of AI through your final poster and oral presentation — and test your knowledge in the midterm exam.

🖼️ Activity: AI Poster & Oral Presentation

This is your moment to shine! Each group presents their AI project with clarity and confidence. Inspire us.

📝 Activity: Midterm Exam

A checkpoint to reflect on everything you’ve learned so far. Think critically, stay sharp — you’ve got this.

🚀 Good luck, and see you in the second half of the semester. 🔥

📅 Week 10: Ultralytics | Revolutionizing the World of Vision AI

🎯 Goal: Explore different Computer Vision tasks using YOLOv8 models (n, s, m, l, x), including detection, segmentation, and pose estimation. Students will gain hands-on experience training, evaluating, and improving models on multiple datasets.

🛠️ Activity: Hands-On Labs in Class

Students will run 5 in-class notebooks, covering a variety of vision tasks:

🔗 Files:

🕐 Note: Students are expected to explore all 5 labs during class, experiment with different YOLO families, try augmentation and fine-tuning techniques, and compare model performance across tasks.

💡 Reminder: Focus on learning by doing. Experiment, visualize predictions, and document your results. Bonus points for creative tricks that outperform larger models!

📅 Week 11: Generative AI

🎯 Goal: Explore state-of-the-art Generative AI models across multiple modalities (text, image, video, music, 3D) to understand model behavior, generate outputs from your own prompts, and analyze performance.

🛠️ Lab Activity: Hands-On with Generative AI Models

🔗 Files:

💡 Task: Experiment with the following generative AI tasks:
🖼️ Text ➡️ Image (e.g., Stable Diffusion) – generate at least 3 creative images
🔄 Image ➡️ Image (Style Transfer / Diffusion) – generate at least 2 style variations
✍️ Text ➡️ Text (e.g., GPT-2, GPT-Neo) – ask at least 10 questions/prompts and analyze responses
🖼️ Image ➡️ Text (Captioning with BLIP or similar) – generate captions for at least 5 images

⚡ Extra Credit: Discover newer or more impressive Hugging Face models than the ones provided. Document all outputs, tricks, prompt engineering, and model settings in your report.

📚 Reference and Credit

🙌 Acknowledgments

This course is inspired by the works of CS231n, and the open-source AI community including Hugging Face and OpenAI.

Made with ❤️ for the next generation of Thai AI innovators 🇹🇭

Support Ukraine

This site is open source. Improve this page.