Our 2023 ICASSP paper is now up on arXiv:
Dataset balancing can hurt model performance
https://t.co/TyawmB1OWK
Dataset balancing works differently than you might assume:
- can cause overfitting;
- doesn’t improve performance on rare classes;
- speeds up training convergence.
This intersection has been a disaster for years. Drivers on Delaney block the box at commute hours and beyond.
@ChrisMarteNYC this is your district.
@LincolnRestler @crystalrhudson your constituents who bike into Manhattan have to navigate this every day.
Was out testing new film equipment. Drivers behaving brutally all over, but none so badly as the chaos at Chrystie & Delancy Streets.
Cranks love to complain about #bikenyc and say drivers don't break laws.
We know that's just not true.
@StreetsblogNYC@bikenewyork@OpenPlans
To show the need for our checklist for ML research, we looked for a study to refute. It took just 1 or 2 hours to find that a model claimed to be 97% accurate is no better than random. My working assumption is that most ML-based science these days is junk. https://t.co/0cNVLRpSn0
Seriously, the map in this article replaces Tunisia with a new arm of the Mediterranean. What map tool even allows you to do that?
https://t.co/iyRFXktros
@HeerJeet I’m curious about how you’re using “WASP” here. Are you claiming that literal white Anglo-Saxon Protestants hold power in the US, or are you using it to mean something else?
E.g., cultural hegemony, assets, or other things held by WASPs; or power held by whites generally.
Our 2023 ICASSP paper is now up on arXiv:
Dataset balancing can hurt model performance
https://t.co/TyawmB1OWK
Dataset balancing works differently than you might assume:
- can cause overfitting;
- doesn’t improve performance on rare classes;
- speeds up training convergence.
We used a partial version of the balancing scheme from the PANN and PSLA/AST papers.
Full balancing (PANN, PSLA/AST) hurts mAP on a held-out set, but partial balancing gives a small boost over baseline:
@Julcuba I also see steel road plates in those photos. They are treacherous for cyclists on a good day because they provide very little traction. A line like that means cyclists will end up with both wheels on a plate and can easily end up falling. When they’re wet they’re a nightmare.