Reddit Machine Learning4h ago|Research & PapersOpinions & Analysis

Additional Experiments During Rebuttal Can Worsen Paper Quality

The author discusses how the pressure to provide additional experiments during the rebuttal process for major AI/ML conferences can often be detrimental to the paper. Reviewers are now expected to find issues, leading to requests for numerous supplementary experiments, even on papers that are otherwise accepted.

💡

Why it matters

This issue highlights the tension between rigorous reviewing and maintaining paper quality, which is crucial for the progress of AI/ML research.

Key Points

  • 1Reviewers feel obligated to find issues, even on otherwise accepted papers
  • 2Requests for 5-10 additional experiments during rebuttal are common
  • 3These experiments are often
  • 4 scenarios that don't meaningfully improve the paper
  • 5Unsuccessful rebuttal experiments can give reviewers a
  • 6 moment

Details

The author notes that in the past, it was common to receive

Like
Save
Read original
Cached
Comments
?

No comments yet

Be the first to comment

AI Curator - Daily AI News Curation

AI Curator

Your AI news assistant

Ask me anything about AI

I can help you understand AI news, trends, and technologies