Are we really making much progress? A worrying analysis of recent neural recommendation approaches

Readings
1

Reproducibility and progress in neural networks…

https://dl.acm.org/doi/abs/10.1145/3298689.3347058

2 Likes
2

Hi Pascal,

thanks for this reference. It’s quite interesting (and depressing? even though reproducibility is said to improve over years, which is encouraging) indeed. I’m not a deep learning expert so I’d like someone else’s opinion on this.

Best,

Arnaud

3

Along the same line, this reminds me of the interview of Joelle Pineau, which @hansijzerman sent me a while ago : https://www.nature.com/articles/d41586-019-03895-5