Moritz Nottebaum

PhD student in Computer Vision.
Research focus in Efficient Inference for Visual Object Tracking and Object Detection

[Matteo Dunnhofer @mdunnhofer]

Since I started working on object tracking I wondered whether the representations of deep nets for videos resembled those of the brain. We recently made a step forward towards the answer.

[Moritz Nottebaum @NottebaumM10908]

@the_shweenz There is a lot of hyperparameter optimization and seed optimization in research. But it depends on the area.
And most reviewers wouldn't count 0.1% improv. as a win for the paper.

[Moritz Nottebaum @NottebaumM10908]

@WangMa70190365 If you could assemble the changes in 5 days, they are not too much...
Additionally, reviewers often ask for more than is neccessary for the paper.

[Moritz Nottebaum @NottebaumM10908]

@demisama_ @Jacklu_me Unforgivable seems to imprint my way.
Similarly, reviewers asking for specific things in a rebuttal, and then becoming completely silent, during my last review process.

[Moritz Nottebaum @NottebaumM10908]

@AvivTamar1 Then you are particularly unlucky.
3/4 papers I reviewed were withdrawn and otherwise I didn't see more than 3 comments per paper.
Also you don't have to consider every part of a verbose discussion to make a decision.

[Moritz Nottebaum @NottebaumM10908]

@gabriberton Nobody is saying it has to be perfect. It never will be. But this year neurips is just a mess.

[Moritz Nottebaum @NottebaumM10908]

@gmishne @noisykaze @NeurIPSConf What if the rejects had no substantial criticism and are very short, thus obviously didn't rly read the paper?

[Moritz Nottebaum @NottebaumM10908]

@Ofirlin I reviewed 4 papers, three already withdrew after initial review. The fourth will also pretty certainly rejected. The range of quality is quite large😅

[Moritz Nottebaum @NottebaumM10908]

@demisama_ @Jacklu_me What if their review was the largest, had the most criticism and was the most curios and got the largest answer? 😂

[Moritz Nottebaum @NottebaumM10908]

@sqx_saqlain Iclr deadline is 25th, notification 24th :D

[Moritz Nottebaum @NottebaumM10908]

@ergo_edge At this point, you just have to hope the AC steps in and looks at the paper.
If reviewers do not react at least roughly properly with arguments, their critique should be dismissed and AC should rely on the other reviewers and the own opinion.

[Moritz Nottebaum @NottebaumM10908]

@LiJiahao14460 @kailas_srt CVPR I think rejected 25 papers due to irresponsible reviews.

[Moritz Nottebaum @NottebaumM10908]

@PKUWZP @emnlpmeeting The main bulk of reviewers should not be people that also submitted. It's additional stress that leads to overly negative ratings and not much Engagement.

[Moritz Nottebaum @NottebaumM10908]

@marcel_hussing Same, maximally verbose and also confusing.
Even AC did his response probably with LLM.

[Moritz Nottebaum @NottebaumM10908]

@VarunChandrase3 The system is horrible. The discussion phase makes no sense if nobody discusses. Submission -> review -> rebuttal -> required reaction by reviewers.
We should keep to this.
The discussion phase just adds more noise.

[Moritz Nottebaum @NottebaumM10908]

@Arnobg32 I think these initial scores have more to do with the reviewers, than with the papers.

[Moritz Nottebaum @NottebaumM10908]

@NeurIPSConf It seems now we can't edit it anymore.

[Moritz Nottebaum @NottebaumM10908]

@NeurIPSConf For how long can we edit our initial rebuttal?

[Moritz Nottebaum @NottebaumM10908]

@odakyildiz @Branchini_Nic @priyald17 @NeurIPSConf I don't think this is the case. I think we have 10k for initial rebuttal plus 10k for the rest of the discussion.