Large image datasets: A pyrrhic win for computer vision? | OpenReview
https://openreview.net/forum?id=s-e2zaAlG3IA detailed investigation of the problematic content of just one of the large image databases commonly used for AI/ML research. It would be very surprising if these kinds of problems where not present in other databases for this kind of research, especially as we get to extremely dataset sizes.
Some press on the fallout of this happening: https://www.theregister.com/2020/07/01/mit_dataset_removed/
Tags
Linked Notes
Related By Tags
- ๐ Eye-catching advances in some AI fields are not real | Science | AAAS
- ๐ The steep cost of capture | ACM Interactions
- ๐ [2111.15366] AI and the Everything in the Whole Wide World Benchmark
- ๐ LAION-400-Million Open Dataset - LAION
- ๐ Get Started ยท Snorkel
- ๐ [1903.03129] SLIDE : In Defense of Smart Algorithms over Hardware Acceleration for Large-Scale Deep Learning Systems
- ๐ ai-tree.pdf
- ๐ [2005.03220] Fractional ridge regression: a fast, interpretable reparameterization of ridge regression
- ๐ Straight to Spam
- ๐ Getting machine learning to production ยท Vicki Boykis
Details
- Revised
- Created
- Edited