Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Consider Adding Further Experiment Groups #52

Open
philswatton opened this issue May 5, 2023 · 0 comments
Open

Consider Adding Further Experiment Groups #52

philswatton opened this issue May 5, 2023 · 0 comments
Labels
enhancement New feature or request

Comments

@philswatton
Copy link
Contributor

#33 will soon be resolved with an experiment group that compares untransformed data against transformed data. With data drop combinations and with 3 seeds, this adds an extra 72 dataset pairs to the original 9 for a total of 81.

From the perspective of statistical power, if the correlation between our similarity metrics and transfer attack metrics exists but is small, this may not be enough to avoid a type 2 error. We might therefore wish to add additional experiment groups later on, potentially comparing transformed data against other transformed datasets (or on the more extreme end, adding a new dataset such as MNIST - although this could might a lot of work to make e.g. the configs backwards compatible for replication purposes).

@philswatton philswatton added the enhancement New feature or request label May 5, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

1 participant