Arseny Khakhalin
@khakhalin.bsky.social
Data Scientist in Berlin
Former Bard College prof
For my after-work alter-ego, see @elstersen.bsky.social
Support Ukraine! 🇺🇦
Former Bard College prof
For my after-work alter-ego, see @elstersen.bsky.social
Support Ukraine! 🇺🇦
Pinned
Arseny Khakhalin
@khakhalin.bsky.social
· Apr 14
Minimalistic bifurcation :)
import matplotlib.pyplot as plt
f = lambda x,k,a: (x if k==0 else f(a*x*(1-x), k-1, a))
y = [f(.21, 17 + i % 19, 2.5+1.5*i/10000) for i in range(10000)]
plt.plot(y,'.',markersize=1);
import matplotlib.pyplot as plt
f = lambda x,k,a: (x if k==0 else f(a*x*(1-x), k-1, a))
y = [f(.21, 17 + i % 19, 2.5+1.5*i/10000) for i in range(10000)]
plt.plot(y,'.',markersize=1);
"while appearing to have only two legs at any one time"
somehow feels vaguely relatable
somehow feels vaguely relatable
19th century wooden “walking” doll with a wheel of eight legs ~ when costumed, the doll’s long dress would cover most of the leg mechanism, allowing the doll to “walk” when moving, while appearing to have only two legs at any one time www.theriaults.com/events/listi...
November 27, 2025 at 12:29 PM
"while appearing to have only two legs at any one time"
somehow feels vaguely relatable
somehow feels vaguely relatable
Since instagram turned squares into tiktokrectangles, I can't use it anymore. It was such a neat aesthetic thing, and it survived a bit even after the acquisition. Then videos came. And this year, complete departure. Everything comes to an end, I guess
November 27, 2025 at 9:33 AM
Since instagram turned squares into tiktokrectangles, I can't use it anymore. It was such a neat aesthetic thing, and it survived a bit even after the acquisition. Then videos came. And this year, complete departure. Everything comes to an end, I guess
Reposted by Arseny Khakhalin
Nano Banana / Gemini really is amazing -- I gave it a cameraphone image from 2009 and had it upscale, change perspective, and remove the trash can.
You just straight-up couldn't do this in Photoshop, no matter how good you are.
You just straight-up couldn't do this in Photoshop, no matter how good you are.
November 25, 2025 at 7:11 PM
Nano Banana / Gemini really is amazing -- I gave it a cameraphone image from 2009 and had it upscale, change perspective, and remove the trash can.
You just straight-up couldn't do this in Photoshop, no matter how good you are.
You just straight-up couldn't do this in Photoshop, no matter how good you are.
Ugh this is bad. Apparently it's not legally binding yet, but it's a bad news
The European Parliament m voted in favor of banning social media for people under the age of 16.
483 MEPs voted in favor while 92 voted against. If it becomes law, people will likely have to use IDs to log into their profiles, making anonymous accounts impossible.
483 MEPs voted in favor while 92 voted against. If it becomes law, people will likely have to use IDs to log into their profiles, making anonymous accounts impossible.
November 26, 2025 at 7:28 PM
Ugh this is bad. Apparently it's not legally binding yet, but it's a bad news
PSA: if you are experiencing a pile up, close the comments like that
November 26, 2025 at 7:12 PM
PSA: if you are experiencing a pile up, close the comments like that
When explaining the anti-ai sentiment I thought of saying something like "Imagine if the first use of electricity were not light, but something horrible..." Except that it's exactly what happened! Electric light only got mass introduction after 1900. Electric chair, invented in 1881, first used 1890
November 26, 2025 at 8:24 AM
When explaining the anti-ai sentiment I thought of saying something like "Imagine if the first use of electricity were not light, but something horrible..." Except that it's exactly what happened! Electric light only got mass introduction after 1900. Electric chair, invented in 1881, first used 1890
Buying books is very much a voting process, a process of shaping the future, which is almost separate from your reading these books. Quite oddly, buying books has a value of its own
I once asked a bookseller at a large indie store how many people would have to buy a book for it to get the attention of the store buyer and cause an additional order and they said: Three.
I see some book piracy discourse, and, to make a positive argument in favor of buying books, your marginal ability to influence what books get published and support the careers of writers you like is massive compared to most other forms of media.
November 26, 2025 at 7:50 AM
Buying books is very much a voting process, a process of shaping the future, which is almost separate from your reading these books. Quite oddly, buying books has a value of its own
Reposted by Arseny Khakhalin
Study shows AI models trained on normal chest X-rays predict health insurance types, unveiling hidden socioeconomic signals. This challenges the neutrality of medical data and advocates for scrutinizing biases in clinical algorithms. https://arxiv.org/abs/2511.11030
Algorithms Trained on Normal Chest X-rays Can Predict Health Insurance Types
ArXiv link for Algorithms Trained on Normal Chest X-rays Can Predict Health Insurance Types
arxiv.org
November 26, 2025 at 6:02 AM
Study shows AI models trained on normal chest X-rays predict health insurance types, unveiling hidden socioeconomic signals. This challenges the neutrality of medical data and advocates for scrutinizing biases in clinical algorithms. https://arxiv.org/abs/2511.11030
I'm trying to understand combinatorially invariant RL with attention and somehow it breaks my brain 🫤 It seems that half-understanding two concepts makes it impossible to synthesize them into a shared intuition. Like trying to close a container with freesing fingers in winter: impossible
November 26, 2025 at 7:31 AM
I'm trying to understand combinatorially invariant RL with attention and somehow it breaks my brain 🫤 It seems that half-understanding two concepts makes it impossible to synthesize them into a shared intuition. Like trying to close a container with freesing fingers in winter: impossible
Reposted by Arseny Khakhalin
Ilya!!!!
www.dwarkesh.com/p/ilya-sutsk...
www.dwarkesh.com/p/ilya-sutsk...
Ilya Sutskever – We're moving from the age of scaling to the age of research
“These models somehow just generalize dramatically worse than people. It's a very fundamental thing.”
www.dwarkesh.com
November 25, 2025 at 6:07 PM
Ilya!!!!
www.dwarkesh.com/p/ilya-sutsk...
www.dwarkesh.com/p/ilya-sutsk...
Reposted by Arseny Khakhalin
Good thread about dimensionality here
Y’all are reading this paper in the wrong way.
We love to trash dominant hypothesis, but we need to look for evidence against the manifold hypothesis elsewhere:
This elegant work doesn't show neural dynamics are high D, nor that we should stop using PCA
It’s quite the opposite!
(thread)
We love to trash dominant hypothesis, but we need to look for evidence against the manifold hypothesis elsewhere:
This elegant work doesn't show neural dynamics are high D, nor that we should stop using PCA
It’s quite the opposite!
(thread)
“Our findings challenge the conventional focus on low-dimensional coding subspaces as a sufficient framework for understanding neural computations, demonstrating that dimensions previously considered task-irrelevant and accounting for little variance can have a critical role in driving behavior.”
November 25, 2025 at 5:20 PM
Good thread about dimensionality here
Are you less active than you used to be back before the Events?
And what can we do to make this place better?
And what can we do to make this place better?
Before the monstrous transformation of Twitter to X, it was considered normal to be active on academic Twitter. Now, even though we've migrated to Bluesky ("safe space"), I feel that many ppl (including myself) are less active in posting anything. Have you noticed this trend? Or am I imagining this?
November 25, 2025 at 5:53 PM
Are you less active than you used to be back before the Events?
And what can we do to make this place better?
And what can we do to make this place better?
tragic, as llms were useful for recipes already a year ago, if not earlier! Primordial chatbots already wouldn't tell you about how the yard of their aunt smelled when she was about to start cooking your search term. Were able to adjust to your fridge content, and provide required/optional analysis!
November 25, 2025 at 5:07 PM
tragic, as llms were useful for recipes already a year ago, if not earlier! Primordial chatbots already wouldn't tell you about how the yard of their aunt smelled when she was about to start cooking your search term. Were able to adjust to your fridge content, and provide required/optional analysis!
If you want to work for a company that makes this world better, you have three choices:
- find one
- found one
- fund one
Unfortunately (3) is not an option for me, (2) is unlikely, so I'm kinda stuck with (1)
- find one
- found one
- fund one
Unfortunately (3) is not an option for me, (2) is unlikely, so I'm kinda stuck with (1)
November 25, 2025 at 4:59 PM
If you want to work for a company that makes this world better, you have three choices:
- find one
- found one
- fund one
Unfortunately (3) is not an option for me, (2) is unlikely, so I'm kinda stuck with (1)
- find one
- found one
- fund one
Unfortunately (3) is not an option for me, (2) is unlikely, so I'm kinda stuck with (1)
find the probability of a gridlock p_{n, L} ( r ) as a function of rounabout radius r, where n is the number of buses, and L is the length of each of them
November 25, 2025 at 3:22 PM
find the probability of a gridlock p_{n, L} ( r ) as a function of rounabout radius r, where n is the number of buses, and L is the length of each of them
This entire topic is both awesome, accessible (in formulation), and apparently full of unproven conjectures and best results!!
en.wikipedia.org/wiki/Opaque_...
en.wikipedia.org/wiki/Opaque_...
November 25, 2025 at 3:08 PM
This entire topic is both awesome, accessible (in formulation), and apparently full of unproven conjectures and best results!!
en.wikipedia.org/wiki/Opaque_...
en.wikipedia.org/wiki/Opaque_...
Reposted by Arseny Khakhalin
another thing I like is asking image models 'A cyclops wearing sunglasses'.
This is Nano Banana, the "ultra-realistic" image generation model, responding to that prompt
This is Nano Banana, the "ultra-realistic" image generation model, responding to that prompt
November 25, 2025 at 2:00 PM
another thing I like is asking image models 'A cyclops wearing sunglasses'.
This is Nano Banana, the "ultra-realistic" image generation model, responding to that prompt
This is Nano Banana, the "ultra-realistic" image generation model, responding to that prompt
Reposted by Arseny Khakhalin
Decades after it was raised as an issue in neurophysiology it is nice to see this topic confronted in the ML/neural net community. I just wish the intellectual connections were stronger and more explicit, because we can't sensibly separate ideas from neuro with those in AI.
📍Excited to share that our paper was selected as a Spotlight at #NeurIPS2025!
arxiv.org/pdf/2410.03972
It started from a question I kept running into:
When do RNNs trained on the same task converge/diverge in their solutions?
🧵⬇️
arxiv.org/pdf/2410.03972
It started from a question I kept running into:
When do RNNs trained on the same task converge/diverge in their solutions?
🧵⬇️
November 24, 2025 at 5:10 PM
Decades after it was raised as an issue in neurophysiology it is nice to see this topic confronted in the ML/neural net community. I just wish the intellectual connections were stronger and more explicit, because we can't sensibly separate ideas from neuro with those in AI.
Reposted by Arseny Khakhalin
November 22, 2025 at 3:12 PM
Reposted by Arseny Khakhalin
S-Values are much more interpretable than P-values, yet adoption seems near impossible. I wonder what it would take to make the leap? #statssky #episky #rstats #statistics
October 10, 2025 at 3:50 PM
S-Values are much more interpretable than P-values, yet adoption seems near impossible. I wonder what it would take to make the leap? #statssky #episky #rstats #statistics
Thank math for llms!
> The joke about "what's his beard's name" is a double-entendre that plays on the word "beard." In slang, a "beard" also refers to a person (often a woman) who is used to conceal someone's sexual orientation, commonly in the context of a gay man appearing heterosexual.
til!
> The joke about "what's his beard's name" is a double-entendre that plays on the word "beard." In slang, a "beard" also refers to a person (often a woman) who is used to conceal someone's sexual orientation, commonly in the context of a gay man appearing heterosexual.
til!
Deeply unsettled by this older photo of Mike Johnson with a beard.
November 25, 2025 at 2:31 PM
Thank math for llms!
> The joke about "what's his beard's name" is a double-entendre that plays on the word "beard." In slang, a "beard" also refers to a person (often a woman) who is used to conceal someone's sexual orientation, commonly in the context of a gay man appearing heterosexual.
til!
> The joke about "what's his beard's name" is a double-entendre that plays on the word "beard." In slang, a "beard" also refers to a person (often a woman) who is used to conceal someone's sexual orientation, commonly in the context of a gay man appearing heterosexual.
til!
Reposted by Arseny Khakhalin
Spotify’s SongDNA feature – revealing the musical lineage of tracks – feels profoundly important. It’s a bet on context, understanding music as a conversation across time and artists... 🧵
www.engadget.com/entertainmen...
www.engadget.com/entertainmen...
Spotify's SongDNA feature will show you which songs are sampled on a track
Spotify has acquired WhoSampled, a community-run database that tracks samples, covers, remixes and more.
www.engadget.com
November 25, 2025 at 1:39 AM
Spotify’s SongDNA feature – revealing the musical lineage of tracks – feels profoundly important. It’s a bet on context, understanding music as a conversation across time and artists... 🧵
www.engadget.com/entertainmen...
www.engadget.com/entertainmen...
Automated cars don't solve car dependency, they exacerbate it.
Waymo privatized another public street:
Chanel approaching 4th, San Francisco
Possibly queued for a Billie Eilish show at Chase Center ~half mile away.
The light rail train on 4th seen passing in front of this roboherd has more passenger capacity than all of them combined.
OP: .tiktok.renaspam18
Chanel approaching 4th, San Francisco
Possibly queued for a Billie Eilish show at Chase Center ~half mile away.
The light rail train on 4th seen passing in front of this roboherd has more passenger capacity than all of them combined.
OP: .tiktok.renaspam18
November 25, 2025 at 9:53 AM
Automated cars don't solve car dependency, they exacerbate it.
My idea of Heaven (like, when you have all the time you want, good health, no obligations, no guilt, and no FOMO) - Heaven is taking courses like that. That's Heaven.
I'm teaching a new class at Berkeley next semester:
Linguistics 265: Biological and Artificial Language
From the syllabus:
This class is an introduction to a novel approach to language. Linguistics has predominantly focused on
human language.
Linguistics 265: Biological and Artificial Language
From the syllabus:
This class is an introduction to a novel approach to language. Linguistics has predominantly focused on
human language.
November 24, 2025 at 6:58 PM
My idea of Heaven (like, when you have all the time you want, good health, no obligations, no guilt, and no FOMO) - Heaven is taking courses like that. That's Heaven.
Speaking of Berlin, I discovered this Horror recently.
A ladder of short parallel streets, connectiong Meteor street to Northern Lights street. The streets are called, in order, going Northward:
East
Mercury
Venus
Saturn (!)
Mars (!!)
Jupyter
West
Uranus
Why?? What sort of Borgean list is it?
A ladder of short parallel streets, connectiong Meteor street to Northern Lights street. The streets are called, in order, going Northward:
East
Mercury
Venus
Saturn (!)
Mars (!!)
Jupyter
West
Uranus
Why?? What sort of Borgean list is it?
November 24, 2025 at 6:07 PM
Speaking of Berlin, I discovered this Horror recently.
A ladder of short parallel streets, connectiong Meteor street to Northern Lights street. The streets are called, in order, going Northward:
East
Mercury
Venus
Saturn (!)
Mars (!!)
Jupyter
West
Uranus
Why?? What sort of Borgean list is it?
A ladder of short parallel streets, connectiong Meteor street to Northern Lights street. The streets are called, in order, going Northward:
East
Mercury
Venus
Saturn (!)
Mars (!!)
Jupyter
West
Uranus
Why?? What sort of Borgean list is it?