Inquire about the database split

[Shuyue Jia]

Dear Wei,

Thanks a lot for your excellent summary! It helps a lot in the field.

Wei, I can see that most SOTA works utilized a strategy of 80% of the database is for training and the remaining 20% for testing. Have u ever seen papers with 60% train - 20% validation - 20% testing settings?

If u ever read that papers, can u send me the paper titles?
Thanks in advance, and have a nice day!

Best regards,

Shuyue
Sep 10, 2021

[Wei Zhou]

Dear Wei,

Thanks a lot for your excellent summary! It helps a lot in the field.

Wei, I can see that most SOTA works utilized a strategy of 80% of the database is for training and the remaining 20% for testing. Have u ever seen papers with 60% train - 20% validation - 20% testing settings?

If u ever read that papers, can u send me the paper titles?
Thanks in advance, and have a nice day!

Best regards,

Shuyue
Sep 10, 2021

Hi Shuyue,

The paper entitled "No-reference Image Quality Assessment with Deep Convolutional Neural Networks" used the 60-20-20 setting in the experiments.

Thanks,
Wei

[Shuyue Jia]

Dear Wei,
Thanks a lot for your excellent summary! It helps a lot in the field.
Wei, I can see that most SOTA works utilized a strategy of 80% of the database is for training and the remaining 20% for testing. Have u ever seen papers with 60% train - 20% validation - 20% testing settings?
If u ever read that papers, can u send me the paper titles?
Thanks in advance, and have a nice day!
Best regards,
Shuyue
Sep 10, 2021

Hi Shuyue,

The paper entitled "No-reference Image Quality Assessment with Deep Convolutional Neural Networks" used the 60-20-20 setting in the experiments.

Thanks,
Wei

Dear Wei,

Noted with thanks!

Best regards,

Shuyue
Sep 12, 2021