neural network when n is very large o(n^2) featrue input layer output layer hidden layer 维数 s j + 1 × ( s j + 1 ) s_{j+1} \times (s_j +1) sj+1×(sj+1) forward propagation(前向传播) z(j)=Θ(j−1)a(j−1) AND OR 计算特征 用向量来表示y(i) 编程作业是5000*400 训练完是10 × \times ×n+1个 θ \theta θ max可以返回最大值和索引 [M,I]