tabular data/ noisy instances

LARS-research / S2E

Q. Yao, H. Yang, B. Han, G. Niu, J. Kwok. Searching to Exploit Memorization Effect in Learning from Noisy Labels. ICML 2020

21 stars 3 forks source link

Open nazaretl opened 2 years ago

nazaretl commented 2 years ago

Hi, thanks for sharing your implementation. I have two questions about it:

Does it also work on tabular data?
Is it possible to identify the noisy instances (return the noisy IDs or the clean set)?

Thanks!

quanmingyao commented 2 years ago

Does it also work on tabular data?
- Yes, if you also use over-parameterized networks. You can have a check over Figure 3 in Simplify and Robustify Negative Sampling for Implicit Collaborative Filtering. NeurIPS 2020
Is it possible to identify the noisy instances (return the noisy IDs or the clean set)?
- Do you mean noisy feature? The idea does not ensure this, can empirical may work for very noisy features.

nazaretl commented 2 years ago

thank you! by noisy instances I mean corrupted labels: whether it is possible to find which labels are corrupted?