# CNN architectures and transfer learning

Subject: Deep Learning

Exam: MT2

Reader: https://deep-learning-study-room.vercel.app/#mt2/read/8

Canonical: https://deep-learning-study-room.vercel.app/read/dl/mt2/chapter/8/

AlexNet, Inception, ResNet, bottlenecks and transfer learning. Study their structure, design purpose and calculations.

1. [AlexNet and the reason for its components](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s0/)
2. [Inception examines several spatial scales](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s1/)
3. [Why one by one convolution saves work](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s2/)
4. [Drill 37 Bottleneck cost with new values](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c17-s37/)
5. [ResNet adds a shortcut](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s3/)
6. [Drill 38 Residual addition](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c17-s38/)
7. [Drill 45 Projection shortcut](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c17-s45/)
8. [Transfer learning from a small dataset](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s4/)
9. [Inception concatenation and residual addition](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s7/)
10. [Drill 44 Inception concatenation and parameter count](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c17-s44/)
11. [Applications of convolutional neural networks](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s9/)
12. [Reading LeNet and AlexNet layer diagrams](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s10/)
13. [Top-1 and top-5 error: what an ImageNet result means](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s11/)
14. [VGG16: small filters in a regular sequence](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s12/)
15. [Global average pooling: one average for each channel](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s13/)
16. [The faculty's 480-channel bottleneck numerical](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s14/)
17. [Batch normalization inside a residual network](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s15/)
18. [Residual-block order and architecture qualifications](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s16/)
19. [Choosing frozen layers from dataset size and similarity](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s17/)
20. [Four ways to reuse a pretrained model](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c8-s18/)
21. [F25 Q2b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c11-s2/)
22. [F25 Q3b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c11-s3/)
23. [R26 Q2b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c12-s2/)
24. [R26 Q7b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c12-s8/)
25. [F23 Q4a](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c13-s3/)
26. [F23 Q5a](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c13-s4/)
27. [F23 Q6b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c13-s5/)
28. [R24 Q6b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c14-s6/)
29. [F22 Q4a](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c15-s5/)
30. [R23 Q4b](https://deep-learning-study-room.vercel.app/read/dl/mt2/section/mt2-c16-s6/)