AI & Machine Learning
Build a model. Test its limits.
Students train, evaluate and break their own machine-learning models, then explain where the model fails.
Illustrative image: CC0 / rawpixel
- Format
- Short course
- Length
- 10 sessions × 90 min
- Ages
- S3–S6
- Group size
- 12–24 students
What students do
- Collect and clean a small dataset
- Train and evaluate a classifier
- Identify bias and overfitting
- Deploy a model behind a simple interface
Tools
- Python
- scikit-learn
- Teachable Machine
- Jupyter
Students collect a small dataset, train a classifier, and then try to make it fail. The interesting work is in the error analysis: which examples are wrong, and why.
The course ends with a working demo and a short written explanation of the model’s limits. No black boxes.
This programme can be adapted for different ages, session lengths and group sizes. Ask for a proposal with a session-by-session outline.
Language: English or Cantonese
Location: Your school or M21, Aberdeen
How it runs
Brief
We agree the cohort, timetable and what a finished project looks like.
Build
Students work in teams on a real brief, with a taught input at the start of each session.
Test
Teams measure against constraints and document what failed and why.
Show
A working demo, a short presentation and a record of each student’s contribution.
Related programmes
AI & software · S3–S6
Ask about AI & Machine Learning — dates, cohort size, fees and a written outline.
hello@badpig.hk