Loss functions (MSELoss, CrossEntropyLoss) in PyTorch - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is the purpose of a loss function in machine learning?

A loss function measures how well the model's predictions match the true values. It helps the model learn by showing how much it is wrong.

beginner

What does MSELoss stand for and when is it used?

MSELoss stands for Mean Squared Error Loss. It is used for regression tasks where the goal is to predict continuous values.

beginner

What type of problems is CrossEntropyLoss used for?

CrossEntropyLoss is used for classification problems, especially when predicting categories or classes.

intermediate

How does MSELoss calculate the error?

MSELoss calculates the average of the squares of the differences between predicted and true values. Squaring makes bigger errors count more.

intermediate

Why is CrossEntropyLoss suitable for classification?

CrossEntropyLoss compares the predicted probabilities with the true class labels using a formula that penalizes wrong predictions more when the model is confident but wrong.

Which loss function is best for predicting house prices?

A Binary Cross Entropy

B MSELoss

C Hinge Loss

D CrossEntropyLoss

MSELoss is used for regression tasks like predicting house prices, which are continuous values.

CrossEntropyLoss is mainly used for:

A Classification problems

B Regression problems

C Clustering problems

D Dimensionality reduction

CrossEntropyLoss is designed for classification tasks where the output is a category.

What does MSELoss penalize more?

A Small errors

B Correct predictions

C Large errors

D Random noise

MSELoss squares the errors, so larger errors have a bigger penalty.

In PyTorch, which loss function expects raw scores (logits) as input?

A MSELoss

B Huber Loss

C L1Loss

D CrossEntropyLoss

CrossEntropyLoss expects raw scores (logits) and applies softmax internally.

Which loss function would you use for a multi-class classification problem?

A CrossEntropyLoss

B MSELoss

C Binary Cross Entropy

D Mean Absolute Error

CrossEntropyLoss is suitable for multi-class classification problems.

Explain the difference between MSELoss and CrossEntropyLoss and when to use each.

Think about the type of output your model predicts: numbers or categories.

Describe how CrossEntropyLoss works with model outputs and true labels.

Consider how the loss handles probabilities and class labels.