Hiroki Naganuma

1.

Screenshot 2021-11-28 at 5 58 02 PM

Backpropを全部記述せよ

Screenshot 2021-11-29 at 5 23 13 PM

2.

Screenshot 2021-11-28 at 5 58 09 PM

Lossをl, f, g で書き換える

3.

Screenshot 2021-11-28 at 5 58 17 PM

線形で簡単に考える場合

Screenshot 2021-11-28 at 5 33 39 PM

ヤコビアンで考える

Screenshot 2021-11-28 at 5 34 58 PM Screenshot 2021-11-28 at 6 00 45 PM

4.

Taking advantage of the above, rewrite the gradient in a form that highlights the exponentially decaying or exploding effects known as gradient vanishing and gradient explosion. What condition on the eigenvalues of per-time step Jacobians is guaranteed to give rise to gradient vanishing?