The extra validation set is useless: cross-validation can pick both your model and your number of…
📰 Medium · Data Science
Learn why a separate validation set may not be necessary for early stopping and how cross-validation can be used instead
Action Steps
- Read the original Medium post to understand the heretical claim
- Apply cross-validation to pick both your model and the number of parameters
- Configure your model development workflow to use cross-validation for early stopping
- Test the performance of your model using cross-validation
- Compare the results with and without a separate validation set
Who Needs to Know This
Data scientists and machine learning engineers can benefit from this insight to optimize their model development workflow
Key Insight
💡 Cross-validation can be used for early stopping, eliminating the need for a separate validation set
Share This
💡 Cross-validation can pick both your model and number of parameters, making a separate validation set unnecessary #MachineLearning #DataScience
Key Takeaways
Learn why a separate validation set may not be necessary for early stopping and how cross-validation can be used instead
Full Article
A few years ago I wrote a short Medium post with a slightly heretical claim: you don’t need a separate held-out set for early stopping (to… Continue reading on Medium »
DeepCamp AI