理学 随机梯度下降 释 stochastic gradient descent; SGD 随机梯度下降 依据损失函数的梯度，在一次迭代中对每一个训练实例进行预测，并重复迭代该过程到一定的次数，从而最小化一个函数的过程。在机器学习中，可以利用该方法来最小化训练模型中的误差，即每次迭代时完成一次评估和更新。
人工智能 神经网络 释文 stochastic gradient descent 随机梯度下降法
- backup to tape, B2T
- backup window
- Backus
- Backus-Naur form
- Backus-Naur form, BNF
- Backus normal form
- back value
- back vowel
- backwall
- backward
- backward algorithm
- backward analysis
- backward Brillouin scattering
- backward chained reasoning
- backward chained reasoning, backward reasoning
- backward channel
- backward crosstalk
- backward derivative
- backward derivative function
- backward difference
- backward difference operator
- backward difference quotient
- backward differentiability
- backward differential equation
- backward differentiation formula