[PDG 457] Not All Explanations for DL Phenomena Are Equally Valuable

Link to article: https://arxiv.org/pdf/2506.23286
Title: Not All Explanations for Deep Learning Phenomena Are Equally Valuable
Content: This position paper argues that many surprising deep learning phenomena studied in recent research (like double descent, grokking, and the lottery ticket hypothesis) have limited evidence of appearing in real-world applications and shouldn't be treated as isolated puzzles requiring bespoke explanations. Instead, the authors propose that these phenomena should be valued as unique settings for refining broader theoretical understanding of deep learning principles, with research efforts aligned toward practical progress in the field rather than phenomenon-specific resolutions.
Slack link: ml-ka.slack.com, channel: #pdg. Please join us -- if you cannot join, please message us here or to mlpaperdiscussiongroupka@gmail.com.

In the Paper Discussion Group (PDG) we discuss recent and fundamental papers in the area of machine learning on a weekly basis. If you are interested, please read the paper beforehand and join us for the discussion. If you have not fully understood the paper, you can still participate – everyone is welcome! You can join the discussion or simply listen in. The discussion is in German or English depending on the participants.