search

siyuhuang / TOD

PyTorch Implementation of Temporal Output Discrepancy for Active Learning, ICCV 2021
MIT License
39 stars 4 forks source link

A related paper #1

Closed kleinzcy closed 3 years ago

kleinzcy commented 3 years ago

Hi, Siyu Huang:

Thanks for your nice work. I learned a lot from your theoretical analysis in part 3. After carefully reading your paper, I found there is a miss of ECCV 2020 paper which is similar to your work. Both borrow ideas from semi-supervised learning. What's more, I am curious about the comparison and analysis between your work and the above ECCV paper under the same setting.

Best wishes.

siyuhuang commented 3 years ago

Hi @kleinzcy , thank you for your interest in our work.

I guess you mean the paper "Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Cost". Its idea is adding data augmentations to samples then measuring the consistency. Samples with larger inconsistency will be selected for oracle annotation. The TOD of this paper is a temporal consistency method that measures the consistency between model outputs at different time steps, given the same sample.

In summary, the ECCV20 paper utilizes the augmentation-based consistency, and TOD utilizes the temporal consistency. Hope this could address your comment. Also, their code is not publicly available, so we did not include it in our experiments.

kleinzcy commented 3 years ago

Thanks for your reply. Your answer solves my concern. Thanks!