深度学习之梯度下降
梯度下降θ∗\theta^*θ∗=argminθ\arg min_\thetaargminθL(θ∗\theta^*θ∗)xuexiL:损失函数 θ∗\theta^*θ∗:参数现在假设θ\thetaθ有两个变量,分别为{θ1\theta_1θ1,θ2\theta_2θ2}随机设定θ0\theta^0θ0=(θ10θ20)\begin{pmatrix}\theta^0_1\\\theta^0_2 \end{pmatrix}(θ10θ20)∇
abla∇L(θ\theta
abla∇L(θ\theta




