COMPSCI 714 — AI Architecture and Design: Exam Killer Book
How to Get an A in One Day (Feynman Method)
This book is reverse-engineered from every past exam (2024, 2025, Practice) with official marking schemes. Every concept is ranked by exam frequency.
Your One-Day Battle Plan (Feynman Whiteboard Method)
The Feynman Technique: Grab a blank sheet. Write the topic. Explain it out loud as if teaching a 12-year-old. When you get stuck, that's your gap. Go back, learn it, explain again.
Morning (3 hrs) — Build Understanding
| Time | Action | What to Do |
|---|---|---|
| 9:00-9:30 | Read Part 0 | Skim exam analysis + frequency map. Know what's coming. |
| 9:30-10:30 | Whiteboard Session 1 | For each MUST topic, read only the Feynman Draft. Close book. Grab paper. Talk out loud. Draw diagrams. Write what you know. Find your gaps. |
| 10:30-11:30 | Whiteboard Session 2 | Read formal sections for your gaps. Close book. Re-explain. Repeat until you can explain CNN calculations, bias-variance diagnosis, and transformer architecture from memory. |
| 11:30-12:00 | CNN Drill | Do 3 CNN dimension calculations by hand. This WILL be on the exam. |
Afternoon (3 hrs) — Practice Exam Questions
| Time | Action | What to Do |
|---|---|---|
| 13:00-13:55 | Mock Exam 1 | Time yourself. 55 minutes. No book. Simulate real conditions. |
| 13:55-14:30 | Check answers | Compare with answer key. Mark your weak spots. |
| 14:30-15:25 | Mock Exam 2 | Another timed attempt. |
| 15:25-16:00 | Review gaps | Re-read chapters for any remaining weak spots. |
Evening (2 hrs) — Cheat Sheet + Final Review
| Time | Action | What to Do |
|---|---|---|
| 19:00-20:00 | Make cheat sheet | Double-sided A4 handwritten (exam allows this!) |
| 20:00-21:00 | Final Feynman pass | Walk around. Explain each MUST topic out loud. No notes. |
What to Put on Your Cheat Sheet
Side 1 — Formulas & Calculations:
CNN CONV output: floor((n + 2p - f) / s) + 1
CNN POOL output: floor((n - f) / s) + 1
Valid padding: p = 0 Same padding: output = input size
Accuracy = (TP + TN) / (TP + TN + FP + FN)
Precision = TP / (TP + FP)
Recall = TP / (TP + FN)
F1 = 2 * P * R / (P + R)
Side 2 — Decision Trees & Key Points:
DIAGNOSIS FLOWCHART:
Train HIGH, Val LOW → Overfitting (high variance)
→ Fix: regularisation, more data, data augmentation, smaller model
Train LOW, Val LOW → Underfitting (high bias)
→ Fix: bigger model, more features, train longer, remove regularisation
Train HIGH, Val HIGH → Good fit!
OUTPUT ACTIVATION:
Multi-class (one label) → Softmax
Multi-label (many labels) → Sigmoid
Regression → Linear (no activation)
BATCH NORM EFFECTS: faster training, reduce vanishing gradients,
regularisation effect, less sensitive to weight init
Exam Format
| Detail | 2025 | 2024 |
|---|---|---|
| Time | 60 min (5 read + 55 write) | 60 min (5 read + 55 write) |
| Marks | 20 | 30 |
| Questions | 6 short-answer | 7 short-answer |
| Allowed | Double-sided handwritten notes | Double-sided page of notes |
Golden rule: "Quality over quantity" — be concise. A 3-sentence precise answer beats a full-page ramble.
考前心理建设（Mental Preparation）
作为中国留学生，你的 ML 概念理解可能比很多本地学生都强。你唯一需要练的是：
- 先说结论（不要铺垫）
- 用题目的数字（不要泛泛而谈）
- 连接词（because, therefore, however — 让逻辑清晰）
- 不要怕犯语法错误（内容正确比语法完美重要100倍）
记住：考官打分看的是你理解不理解，不是你英语好不好。 一个有小语法错误但逻辑清晰的答案 >> 一个语法完美但空洞的答案。