ddofer.bsky.social
@ddofer.bsky.social
I’m giving a talk at @PyDataTLV !
“Is This Feature Actually Interesting? ML & LLMs for Automated Insight Discovery”
cfp.pydata.org/pydata-tel-a...

I'll show how to automatically find "Interesting" features in data, and what "interesting" means!
Paper: arxiv.org/abs/2505.13534
#pydata
Is This Feature Actually Interesting? ML & LLMs for Automated Insight Discovery (EN) PyData Tel Aviv 2025
Machine learning excels at prediction, but often leaves data scientists manually sifting through feature importance lists to find truly interesting insights. This talk introduces "InterFeat," an autom...
cfp.pydata.org
November 3, 2025 at 9:05 PM
Heading to London @mcmcomiccon.bsky.social - Hit me up for #Photography or just geek / AI chatting!
October 17, 2025 at 10:46 AM
This goes out to the cat people
September 26, 2025 at 4:14 PM
Reposted
Dutch late night TV has its take
September 19, 2025 at 2:39 PM
'Almost like science fiction': European ant is the first known animal to clone members of another species www.livescience.com/animals/ants...

- Something fun for a change. This is some weird-ass-awesome biology!
'Almost like science fiction': European ant is the first known animal to clone members of another species
A species of ant found scurrying across southern Europe is the first animal found that clones males of another species.
www.livescience.com
September 18, 2025 at 5:32 AM
Kinda weird that I need different prompts to GPT5 to get it to proofread my manuscript properly. Each iteration + prompt yields different mistakes. And I'm looking for basic mistakes (e.g., grammar/spelling level)!
This is a super common usecase, i'm surprised at the variation
#writing #gpt #ai
September 4, 2025 at 10:38 AM
1/ Our paper “Protein Language Models Expose Viral Immune Mimicry” is now published in Viruses!
We show that protein Language Models can identify viral proteins, and those that fool our immune system.
DOI: 10.3390/v17091199.
www.mdpi.com
September 1, 2025 at 11:19 AM
Love this graph for (science) problem-picking. Credit: @urialonlab.bsky.social
August 25, 2025 at 8:15 AM
Reposted
The person who says "oh I read Philosopher's Stone and it just didn't click" is probably telling the truth.

The person who says "I KNEW JK ROWLING WAS SCUM THE MOMENT SHE CALLED VERNON FAT" absolutely wrote a Library of Alexandria's worth of Harry/Draco fanfiction.
August 16, 2025 at 12:27 AM
Tl;DRs for reading research papers:
“What’s the problem they’re trying to solve?”, “How do they solve it?”, “Does it work?”, "Does it smell fishy?"
August 5, 2025 at 7:45 AM
Back from 6 weeks travel in Asia.
Turns out my camera had a big dust spot on the lens. Sigh.
July 8, 2025 at 3:07 PM
Our newest paper is out!
InterFeat: An Automated Pipeline for Finding Interesting Hypotheses in Structured Biomedical Data
tl;dr: It's an automated method using AI/LLMs to find interesting features in data. Doctors reviewed them on different diseases and the UK BioBank.
May 21, 2025 at 9:26 AM
"Many things may change the nature of a man. Whether regret, or love, or revenge or fear - whatever you believe can change the nature of a man, can. I've seen belief move cities, make men stave off death, and turn an evil hag's heart half-circle[...]" www.goodreads.com/quotes/542497 via #goodreads
A quote by Chris Avellone
You are wrong. If there is anything I have learned in my travels across the Planes, it is that many things may change the nature of a man. Whether regret...
www.goodreads.com
May 7, 2025 at 8:22 AM
Welp, no ICML for me this year :(
May 2, 2025 at 10:08 AM
Reposted
People will always say they’re your friend and they support you no matter what but then they’ll say some toxic shit to you like “you have to let go of the cursed amulet, it’s changing who you are” and “don’t let its power corrupt you” all because they want the amulet for themselves
April 10, 2025 at 4:54 PM
Reposted
FUNDSTRAT tonight:

“.. in the last few days, we have had many conversations with macro fund managers. .. A few have quietly wondered if the President might be insane.”
April 9, 2025 at 2:37 AM
#ICML Reviews are out. Damn that reviewer #2 :(
(Who also gets flagged as AI generated by 3 seperate tools. But kinda sus if I complain about them)
March 25, 2025 at 9:40 AM
Venice: Masks
March 10, 2025 at 10:38 AM
www.lesswrong.com/posts/DfrSZa...

The interesting stuff we could do if not for ethics committees, (and off site edit mistakes)
How to Make Superbabies — LessWrong
EDIT: Read a summary of this post on Twitter …
www.lesswrong.com
March 10, 2025 at 10:35 AM
Reposted
when my family asks me about the impact of my research
March 4, 2025 at 5:34 PM
Paper submitted! Now to wait for after the review period so I can share. (It's really cool) #academia #research #PhD
January 31, 2025 at 12:02 PM
And here I thought the 3AM rockets would be the nasty highlight of the day. 🤮 (Gaiman article).
January 14, 2025 at 8:57 AM