Even in machine learning, as important as knowing how to listen to the data is knowing when to ignore it.

May 20, 2022 · 5:17 PM UTC

16
41
3
338
Replying to @gdb
Reminds me of Occam Razor (though not exactly equal)
1
1
Replying to @gdb
More data can hurts sometime arxiv.org/abs/1912.02292

Deep Double Descent: Where Bigger Models and More Data Hurt

We show that a variety of modern deep learning tasks exhibit a "double-descent" phenomenon where, as we increase model size, performance first gets worse and then gets better. Moreover, we show...

arxiv.org
1
Replying to @gdb
"The result of this experiment makes me think I have bias/noise in my data generation/collection process" related arxiv.org/pdf/2006.04730.pdf
1
Replying to @gdb
and when to counter it … with truth!
Dr. Novo
 @novocrypto
23 May 2022
I asked AI assistant @OpenAI to pretend to be @VitalikButerin while I pretend to be his 5 year old kid! See what I got 🤯 Deeply saddened by the GENDER BIAS in GPT3 AI assistant. This shit needs to be addressed before GPT3 models creeping in our lives. @elonmusk @GirlsWhoCode
Replying to @gdb
Garbage in, garbage out. Ignore the garbage.
Replying to @gdb
It’s called feature engineering ;)
Replying to @gdb
Knowing which datapoints can be safely ignored is the hard part! Here’s an open-source library that helps you identify *which* datapoints deserve your attention and may require some manual triage: github.com/cleanlab/cleanlab…

GitHub - cleanlab/cleanlab: The standard data-centric AI package for data quality and machine...

The standard data-centric AI package for data quality and machine learning with messy, real-world data and labels. - GitHub - cleanlab/cleanlab: The standard data-centric AI package for data qualit...

github.com
Replying to @gdb
as in “attention is all we need”..
1
Replying to @gdb
Do we ignore the data or we ignore the suggestions we incur from our it?
2
Replying to @gdb
Isn't this just a truism of any discriminator?
Load more