Question at position 20 What is gradient descent?A way to calculate the slope of a loss function A regularization techniqueA method to normalize network activationsAn optimization algorithm to find local minimum of a function

Question

BlackTom AI · Accepted Answer

To understand gradient descent, we need to consider what each option claims and how it aligns with the concept.
Option 1: 'A way to calculate the slope of a loss function.' While gradient relates to the slope (the gradient is the slope vector), gradient descent is not merely calculating the slope. It uses that slope information to adjust parameters iteratively. So this description is incomplete and misleading about the purpose of the method.
Option 2: 'A regularization technique.' Regularization techniques (like L1/L2, dropout) are designed to prevent overfitting by constraining model complexity, whereas gradient descent is a training procedure that updates weights. This option confuses the optimization process with a regularization strategy, making it incorrect.
Option 3: 'A method to normalize network activations.' Normalization methods (such as batch normalization) aim to stabilize activations across layers, not to optimize the objective by moving parameters. Gradient descent does not perform normalization of activations, so this option does not describe gradient descent.
Option 4: 'An optimization algorithm to find local minimum of a function.' This statement accurately captures the essence of gradient descent: it uses the gradient to iteratively move parameters in the direction that reduces the loss, typically converging toward a local minimum. It reflects the standard definition and purpose of the method.

Question at position 20 What is gradient descent?A way to calculate the slope of a loss function A regularization techniqueA method to normalize network activationsAn optimization algorithm to find local minimum of a function

查看解析

登录即可查看完整答案

类似问题

Which of the following statements about gradient descent and learning rate is true?

Which statement is correct?

在梯度下降中如何更新参数？How do we update the parameters in gradient descent?

Which of the following best describes the role of the gradient in gradient descent?

Which statement best describes the use of gradient descent in linear classification?

How does random initialization help in gradient descent methods?

更多留学生实用工具

考试浏览器助手

风格化写作助手

论文查重助手

文献引用助手

课堂转译助手

课堂笔记助手

Quiz搜索助手

学校历年真题

智能刷题助手

智能匹配练习

希望你的学习变得更简单