mstdn.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A general-purpose Mastodon server with a 500 character limit. All languages are welcome.

Administered by:

Server stats:

13K
active users

#microscopy

6 posts4 participants1 post today

It's a bit magical how adding diversity to a training dataset improves the result of the model. I was analyzing one microscopy experiment, and the segmentation model, trained on the same experiment, was doing well on this particular movie but not on others. I added a few more experiments to the training set, and now the model does much better even on experiments outside of the training set.

: -based multi-channel time-tagging module () for democratising single-photon (SP) :

-parallel multiple event tagging precision: 30 ps
-multiple synchronisation event precision: 4 ns
-requires
-cost ~$3000

Article: doi.org/10.1038/s41467-022-350
Web: brighteyes-ttm.readthedocs.io/
GitHub: github.com/VicidominiLab/Brigh

#Vegetative #electron #microscopy” is #nonsense. And yet several scientific #papers have included the phrase over the past few years.

When two papers from the 1950s were digitized, two columns of text may have been read together incorrectly, the “vegetative” from one column smushed up with the “electron microscopy” in the other.

The #researchers discovered the phrase started showing up in responses generated by #OpenAI’s GPT 3.0, which was released in 2020.

freethink.com/artificial-intel

FreethinkAI thinks “vegetative electron microscopy” is real — it’s notThe appearance of a nonsense term in multiple scientific papers has exposed how flawed data in training sets can become embedded in AIs.