Counterintuitive Coin Flips meet Deep Neural Network Theory

What do the Alice HH vs Bob HT coin flip game and deep neural networks have in common? The same mathematical approximation technique, Edgeworth expansions, can be used on both of them!

Link to my note on Edgeworth expansions in Alice HH vs Bob HT problem. This is still work in progress with Svante Janson where we have to prove some technical approximations to prove the Edgeworth series expansion holds in this case: https://publish.obsidian.md/nicam/Fun...

Link to my paper on Edgeworth expansions in Deep Neural Networks: https://openreview.net/forum?id=92q7W...

Chapters:
0:00 Vector Institute Workshop
0:46 Alice HH vs Bob HT Coin Flip Problem
3:43 Comments from Last Video
5:08 Three Coin Flip Example
9:53 Self Overlap of a Sequence Definition
16:08 Edgeworth Expansions with No Extra Terms
17:45 Edgeworth Expansion with One Extra Term
20:38 Edgeworth Expansion with Two Extra Terms
22:04 Edgeworth in Neural Networks

Music Credits:
"A Very Brady Special" Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 4.0 License
http://creativecommons.org/licenses/b...