Update on our similarity-repository 🚨 More than 200 similarity measures across 32 papers are now registered! We'll also be presenting our work as an oral at the @neuripsconf.bsky.social @unireps.bsky.social workshop! (1/6) - ThreadSky

About ThreadSky

Update on our similarity-repository 🚨
More than 200 similarity measures across 32 papers are now registered!

We'll also be presenting our work as an oral at the @neuripsconf.bsky.social
@unireps.bsky.social
workshop! (1/6)

Comments

nacloos.bsky.social•85 days ago

The number of similarity measures keeps growing and comparisons across studies become more and more challenging, mainly due to differences in naming and implementation conventions. (2/6)

nacloos.bsky.social•85 days ago

Standardization is challenging because the field keeps evolving. The ideal naming convention shouldn't be static but should flexibly evolve too!

We propose a package and a procedure as a step towards this goal. (3/6)

nacloos.bsky.social•85 days ago

For example, we applied our procedure to derive a standardized naming convention unifying CKA measures. (4/6)

nacloos.bsky.social•85 days ago

Naming conventions with too few names led to consistency errors when comparing CKA implementations across papers.

We iteratively refined our naming convention to resolve inconsistencies while keeping low naming complexity. (5/6)

nacloos.bsky.social•85 days ago

Our package aims at being exhaustive. If your implementation is missing, checkout our GitHub to add your similarity measures!

GitHub: https://github.com/nacloos/similarity-repository
Paper: https://openreview.net/forum?id=vyRAYoxUuA

Work with Robert Yang and Chris Cueva. (6/6)

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply