Issues
- It's not clear on how many cases to assign to the training and validation sets
- The result might highly depend on the algorithm which is used to split the data into training and validation set
- Notion of "increasing validation error" is ambiguous; it may go up and down numerous times during training. The safest approach is to train to convergence, then determine which iteration had the lowest validation error. This impairs fast training, one of the advantages of early stopping.
Read more about this topic: Early Stopping
Famous quotes containing the word issues:
“I can never bring you to realize the importance of sleeves, the suggestiveness of thumb-nails, or the great issues that may hang from a boot-lace.”
—Sir Arthur Conan Doyle (1859–1930)
“How to attain sufficient clarity of thought to meet the terrifying issues now facing us, before it is too late, is ... important. Of one thing I feel reasonably sure: we can’t stop to discuss whether the table has or hasn’t legs when the house is burning down over our heads. Nor do the classics per se seem to furnish the kind of education which fits people to cope with a fast-changing civilization.”
—Mary Barnett Gilson (1877–?)
“The current flows fast and furious. It issues in a spate of words from the loudspeakers and the politicians. Every day they tell us that we are a free people fighting to defend freedom. That is the current that has whirled the young airman up into the sky and keeps him circulating there among the clouds. Down here, with a roof to cover us and a gasmask handy, it is our business to puncture gasbags and discover the seeds of truth.”
—Virginia Woolf (1882–1941)