Aubin Ramon

speaker

346 appearances 1 recordings 1 series first heard Jul 2026 last heard 6 Jul

Aubin Ramon’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Jul 2026 with 1.

Appearances

newest first · ▶ plays the moment

Talking Techniques · AI & Antibodies miniseries | Nanobody thermostability prediction and the great data challenge · 6 Jul 2026 podcast

It's no linear.

relationship between the performances and the data set sizes.

It will be actually um a plateau.

So at some point more you add more data, you won't get much performances.

And we saw this plateau around one thousand data points.

And we had six hundred, seven hundred data points.

So I don't think uh if you go to thousands of data points you will solve a problem.

What we'll solve a problem is to spend money and amount and effort

non the amount but the quality of the data you're gonna um compile.

So what's more interested is rather than thermosability, which is a decontextualized um property decontextualized from uh the buffer you're gonna use in the environment, the pH.

it would be yeah to to develop this data set which is standardized at different conditions, so then you can have an environment specific for more stability.

And this is much more interesting for engineering.

uh of nanobodies in very precise therapies and castle cells, intra bodies or et cetera.

Um so that would be much more interesting that having a bigger bigger data set.

And then you will need I think to put to spend more time on the quality of equ architecture and the induction biases you want

to know and the one that matters for application.

And for instance, I went to a conference last year, international one, big one Seneop was there, presented basically by version of Nanomel.

same kind of ensemble learning.

Showing 181–200 of 346 · page 10 of 18