Your project · 15%

Subtitles

1. Research question and claim
2. Data, labels, and splits
3. Baseline and model
4. Preprocessing and training
5. Metrics and results
6. Code and reproducibility
7. Failures, limitations, and next step

Milestones

• Reproduce the poster pipeline without looking at the code.
• Defend every design choice against one alternative.
• Explain every result, metric, and baseline on the poster.
• Locate the code behind each poster claim.
• State the main failure mode and next decisive experiment.

Possible exam questions

PROJECT · Q1 — What exactly is your research question, and which result answers it?

PROJECT · Q2 — Why did you choose these data, labels, and splits? Where could leakage enter?

PROJECT · Q3 — Why this model and baseline? What would a simpler alternative test?

PROJECT · Q4 — Which preprocessing, training, and hyperparameter choices most affect the result?

PROJECT · Q5 — What does each reported metric measure, and what do your numbers actually establish?

PROJECT · Q6 — How does the code reproduce each poster figure and table from the raw inputs?

PROJECT · Q7 — Where does your approach fail, what is its strongest limitation, and what would you test next?