CODEWITHSUNDEEP
pythontensorflowkerasdeep-learningtraining-data
Eda
Eda

Reputation: 705

Why is my validation loss lower than my training loss?

Why is my validation loss lower than my training loss? I wonder if those curves are acceptable. I don't know if those curves are true or not.

enter image description here enter image description here

Upvotes: 1

Views: 1380

Answers (2)

sentence
sentence

Reputation: 8903

The general rough reason behind this kind of scenario is that your training set is somehow over-complicated compared to the validation set.

Considering that you used deep-learning as a tag, I would even consider the effect of dropout regularization layers. For a theoretical discussion, see this post.

Upvotes: 0

Sara El
Sara El

Reputation: 26

This case will rarely happen in a real-world dataset, however, it is possible to happen, consider, for example, where the validation data set and the training have roughly the same distribution, with the training set being noisier than the validation dataset, in that case, the learning you are showing is possible.

Upvotes: 1

Related Questions