Optimization Fails
local minima
local maxima
saddle point
critical point (gradient = 0)
Gradient
梯度消失
Hessian
在 saddle point 中可以指出更新方向。也许只存在于低维度,高维度就可以解决。
Batch Size
小批次更新频率高,每次更新不够精准,但...
#include <iostream>
using namespace std;
int main() {
int n, m;
cin >> n >> m;
int dp[n + 1];
int suff[n + 2];
dp[n] = 1;
suff[n] = 1;
suff[n + 1] = 0;
for (int ...
A - Gregor and Cryptography
Gregor is learning about RSA cryptography, and although he doesn't understand how RSA works, he is now fascinated with prime numbers and factoring them.
Gregor's favori...