Lecture - Theories of Deep Learning MT25, XV, A few things we missed and a summary

Created: November 18, 2025 | Updated: November 18, 2025 | About these notes

[[Course - Theories of Deep Learning MT25]]^U
Dropout
Skip connections
Tokenisation
- Tokens are probably different in a chat context vs a coding context
How sparse can you make your nets before losing loads of accuracy?
Major omissions
- How well does depth improve generalisation error?

[[Course - Theories of Deep Learning MT25]]^U

(outgoing)
[[Lecture - Theories of Deep Learning MT25, XIII, Autoencoders]]^U

(sim: 0.664)
[[Paper - When and when can deep networks avoid the curse of dimensionality, Poggio (2016)]]^U

(sim: 0.608)
[[Notes - Computer Vision MT25, Dropout]]^U

(sim: 0.6)
[[Lecture - Machine Learning MT23, XIV]]^U

(sim: 0.597)
[[Lecture - Machine Learning MT23, XVI]]^U

(sim: 0.595)