Which statement is correct?

Question

BlackTom AI · Accepted Answer

Question restatement: Which statement is correct?

Option A: 'Stochastic Gradient Descent (SGD) computes the gradients using the whole training set to update the model parameters once.' This is inaccurate. SGD updates parameters using one or a few samples at a time, not the entire training set, which is characteristic of batch methods. The description more closely matches Batch Gradient Descent in its general idea, but SGD specifically uses stochastic updates rather than the full batch.

Option B: 'Batch Gradient Descent (BGD) computes the gradients using one data point to update the models parameters once.' This is a common point of confusion. In BGD, gradients are computed using the entire training set as one batch before updating parameters, not from a single data point. Therefore, this statement is incorrect because it describes SGD behavior rather than BGD.

Option C: 'Mini-batch Gradient Descent has the most bouncing behavior compared to SGD and BGD.' The phrase 'bouncing behavior' is informal and vague here, but in standard optimization terminology, mini-batch gradient descent balances between the noisiness of SGD and the stability of BGD. It does not inherently have the most bouncing behavior; rather, SGD tends to be noisier and BGD the most stable. So this statement is misleading and not accurate in a formal sense.

Option D: '10 training epochs mean each data point has the opportunity to update the model parameters 10 times.' This statement aligns with the definition of an epoch: one pass over the entire training dataset. If there are 10 epochs, each data point is used to update parameters once per epoch, totaling 10 updates per data point across the training run. This matches standard understanding of epochs in training neural networks.

In summary, the incorrect options hinge on common misconceptions: A confuses SGD with full-batch usage, B swaps the roles of SGD and BGD, and C relies on an imprecise notion of 'bouncing' without a formal basis. Option D accurately describes the epoch concept and how multiple epochs translate to multiple updates per data point, which is consistent with how training loops are typically structured.

Which statement is correct? 单项选择题

类似问题

Which if the following answers are correct regarding "Gredient descent"?Select all the correct answers.

Why does the gradient descent algorithm not work for training a linear classifier?

Which parameter determines the size of the improvement step to take on each iteration of Gradient Descent?

Optimization algorithms like gradient descent are typically used in machine learning for which of the following (select all that apply):

The learning rate (α - alpha) as related to gradient descent is best described as:

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

更多留学生实用工具

智能学习助手

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