michaelhoffman.bsky.social
Chair, Computational BIology and Medicine Program, Princess Margaret Cancer Centre, University Health Network.
Associate Professor, Medical Biophysics, University of Toronto.
Disclosures: https://github.com/michaelmhoffman/disclosure/
3,383 posts
15,795 followers
1,538 following
Regular Contributor
Active Commenter
comment in response to
post
comment in response to
post
This looks interesting academic.oup.com/bioinformati...
comment in response to
post
Another way to express these are
• log(y + m) - log(m)
• log(y + 1) - log(m)
or
• log((y+m)/m)
• log((y+1)/m)
A pseudocount does not *just* avoid 0s or nonfinite quantities. It has the effect of flattening out potentially noisy data. A factor of m will flatten a LOT more than 1.
comment in response to
post
"Even under the Trump administration"
comment in response to
post
Yes, the default tip is different ¯\_(ツ)_/¯
comment in response to
post
2/Left: order.online
Right: doordash.com
Surprised that "DoorDash" is cheaper, even before DashPass discount. I thought it was usually the other way around? Anyway, still a tip option here, but I think I have ordered from order.online without one.
comment in response to
post
1/Comparison of ordering from a local restaurant via the "Online Ordering by DoorDash" channel versus the "DoorDash" channel
comment in response to
post
I believe some of those do not have a tip option, and dasher just gets a standard (low) amount.
comment in response to
post
Are you doing this through DoorDash directly or a "white label" service that uses DoorDash under the hood, like order.online?
comment in response to
post
Ah, the "One Weird Trick" approach. Good luck with that, folks
comment in response to
post
As someone who has done "AI" for 17 years, I almost never describe my work that way, and prefer to be more specific.
If someone submits an application on "AI" to my CIHR panel, and it is not easy to figure out what they are actually doing, their app is in trouble. But we get a LOT of AI/ML.
comment in response to
post
Yes, but I would argue that the consequences of that are way too indirect for it to be a problem for you. And outweighed by the benefits.
comment in response to
post
Idiots all the way down
comment in response to
post
I don't see the ethical problem of reframing your work to claim it as, for example, "AI" as long as YOUR work is not problematic or can be used in a problematic way by others.
comment in response to
post
🧐
comment in response to
post
I didn't want to ruin your fun earlier, but I have to admit now Leighton Buzzard isn't named after a bird either. A whole aviary of false etymologies
comment in response to
post
And only 2 h 22 min from Leighton Buzzard
comment in response to
post
womp womp
comment in response to
post
Another position available at McGill in Psychiatry and Single Cell Genomics
can-acn.org/assistant-as...
comment in response to
post
And one more again at McGill in artificial intelligence in Psychiatry
can-acn.org/assistant-pr...
comment in response to
post
Well-argued by @lanrickbennett.bsky.social. Fletcher continues to be a disappointment.
comment in response to
post
ngl lasted way longer than I thought
comment in response to
post
@kieranrcampbell.bsky.social Congratulations!
comment in response to
post
You've never heard it called the Sorbonne of the West? The Oxford of the Americas?
comment in response to
post
Freeze them. Or turn them into jam. Or bake pies.
comment in response to
post
This 1999 paper claims 24, but data not shown. Change could be due to advances in women's health in the last 26 years. improbable.com/airchives/pa...
comment in response to
post
There's always someone credulous enough