Dezhe Jin
@dezhejin.bsky.social
Associate Professor of Physics, research in computational neuroscience, brain inspired AI, and physics. Enjoy gazing the night sky and playing music. #neuroscience #physics #astrophotography #music
Where are neuroscientists hanging around these days? Back to X?
November 20, 2025 at 4:05 AM
Where are neuroscientists hanging around these days? Back to X?
www.nature.com/articles/d41...
Is the academic world waking up against the bean counters? To assess a scientist, read top three papers, consults experts in the field. Not by H-index or total citations. These metrics have been gamed beyond to be useful.
Is the academic world waking up against the bean counters? To assess a scientist, read top three papers, consults experts in the field. Not by H-index or total citations. These metrics have been gamed beyond to be useful.
Introducing the j-metric: a true measure of what matters in academia
Science has become obsessed with publishing numbers. With this new satirical proposal, have we reached peak metric?
www.nature.com
November 19, 2025 at 3:51 PM
www.nature.com/articles/d41...
Is the academic world waking up against the bean counters? To assess a scientist, read top three papers, consults experts in the field. Not by H-index or total citations. These metrics have been gamed beyond to be useful.
Is the academic world waking up against the bean counters? To assess a scientist, read top three papers, consults experts in the field. Not by H-index or total citations. These metrics have been gamed beyond to be useful.
At SFN San Diego, right now, Terry Sejnowski is delivering fascinating talk on Neuro AI. At 78, he has the energy and clarity of a person in his 40s. Amazing scientist.
November 18, 2025 at 5:58 PM
At SFN San Diego, right now, Terry Sejnowski is delivering fascinating talk on Neuro AI. At 78, he has the energy and clarity of a person in his 40s. Amazing scientist.
Reposted by Dezhe Jin
Brain-computer interfaces can currently decode intended language. In future decades, will they decode something closer to thought? What are the medical opportunities and the ethical conundrums? Join guest Sergey Stavisky as we dive into the world of interfacing brains and machines.
November 1, 2025 at 2:02 PM
Brain-computer interfaces can currently decode intended language. In future decades, will they decode something closer to thought? What are the medical opportunities and the ethical conundrums? Join guest Sergey Stavisky as we dive into the world of interfacing brains and machines.
Reposted by Dezhe Jin
Spiking NN fans - the #SNUFA workshop (Nov 5-6) agenda is finalised and online now. Make sure to register (free) soon. (Note you can register for either day and come to both.)
Agenda: snufa.net/2025/
Registration: www.eventbrite.co.uk/e/snufa-2025...
Thanks to all who voted on abstracts!
🤖🧠🧪
Agenda: snufa.net/2025/
Registration: www.eventbrite.co.uk/e/snufa-2025...
Thanks to all who voted on abstracts!
🤖🧠🧪
SNUFA 2025
Spiking Neural networks as Universal Function Approximators
snufa.net
October 23, 2025 at 4:17 PM
Spiking NN fans - the #SNUFA workshop (Nov 5-6) agenda is finalised and online now. Make sure to register (free) soon. (Note you can register for either day and come to both.)
Agenda: snufa.net/2025/
Registration: www.eventbrite.co.uk/e/snufa-2025...
Thanks to all who voted on abstracts!
🤖🧠🧪
Agenda: snufa.net/2025/
Registration: www.eventbrite.co.uk/e/snufa-2025...
Thanks to all who voted on abstracts!
🤖🧠🧪
This event has horrific smell to it.
Breaking News: President Trump pardoned the billionaire founder of the cryptocurrency exchange Binance, Changpeng Zhao. Zhao admitted in 2023 to money-laundering violations that allowed terrorists, drug traffickers and other criminals to move money on the platform. nyti.ms/3J9mazz
October 23, 2025 at 5:33 PM
This event has horrific smell to it.
SGD is much much better than AdamsW for escaping long plateaus in loss function when training transformers. Hard lesson learned after a week of struggles.
October 6, 2025 at 4:40 PM
SGD is much much better than AdamsW for escaping long plateaus in loss function when training transformers. Hard lesson learned after a week of struggles.
Reposted by Dezhe Jin
Interesting paper suggesting a mechanism for why in-context learning happens in LLMs.
They show that LLMs implicitly apply an internal low-rank weight update adjusted by the context. It’s cheap (due to the low-rank) but effective for adapting the model’s behavior.
#MLSky
arxiv.org/abs/2507.16003
They show that LLMs implicitly apply an internal low-rank weight update adjusted by the context. It’s cheap (due to the low-rank) but effective for adapting the model’s behavior.
#MLSky
arxiv.org/abs/2507.16003
Learning without training: The implicit dynamics of in-context learning
One of the most striking features of Large Language Models (LLM) is their ability to learn in context. Namely at inference time an LLM is able to learn new patterns without any additional weight updat...
arxiv.org
October 6, 2025 at 1:30 PM
Interesting paper suggesting a mechanism for why in-context learning happens in LLMs.
They show that LLMs implicitly apply an internal low-rank weight update adjusted by the context. It’s cheap (due to the low-rank) but effective for adapting the model’s behavior.
#MLSky
arxiv.org/abs/2507.16003
They show that LLMs implicitly apply an internal low-rank weight update adjusted by the context. It’s cheap (due to the low-rank) but effective for adapting the model’s behavior.
#MLSky
arxiv.org/abs/2507.16003
Reposted by Dezhe Jin
BREAKING: Mary E. Brunkow, Fred Ramsdell and Shimon Sakaguchi win the Nobel Prize in medicine for work on peripheral immune tolerance.
The Nobel Prize in medicine goes to 3 scientists for work on peripheral immune tolerance
Mary E. Brunkow, Fred Ramsdell, and Shimon Sakaguchi have won the Nobel Prize in medicine for their discoveries about peripheral immune tolerance.
bit.ly
October 6, 2025 at 9:41 AM
BREAKING: Mary E. Brunkow, Fred Ramsdell and Shimon Sakaguchi win the Nobel Prize in medicine for work on peripheral immune tolerance.
Reposted by Dezhe Jin
Exciting news for #drosophila #connectomics and #neuroscience enthusiasts: the Drosophila male central nervous system connectome is now live for exploration. Find out more at the landing page hosted by our Janelia FlyEM collaborators www.janelia.org/project-team....
Male CNS Connectome
A team of researchers has unveiled the complete connectome of a male fruit fly central nervous system —a seamless map of all the neurons in the brain and nerve cord of a single male fruit fly and the ...
www.janelia.org
October 5, 2025 at 3:40 PM
Exciting news for #drosophila #connectomics and #neuroscience enthusiasts: the Drosophila male central nervous system connectome is now live for exploration. Find out more at the landing page hosted by our Janelia FlyEM collaborators www.janelia.org/project-team....
Man, first time seeing a real ADHD, in Statbucks. Sixty-ish old man, cannot sit still for more than 2 seconds. He says hi and good job to the staff frrquency, stands up for something, and says something to anyone passes by. Otherwise looks normal.
September 27, 2025 at 3:15 PM
Man, first time seeing a real ADHD, in Statbucks. Sixty-ish old man, cannot sit still for more than 2 seconds. He says hi and good job to the staff frrquency, stands up for something, and says something to anyone passes by. Otherwise looks normal.
When asked to find a paper on a specific subject, Gemini 2.5 confidently fabricates non-existent paper. Even at this day of AGI talks, AI models hallucinate big time. It can blind side unsuspecting users.
August 20, 2025 at 7:54 PM
When asked to find a paper on a specific subject, Gemini 2.5 confidently fabricates non-existent paper. Even at this day of AGI talks, AI models hallucinate big time. It can blind side unsuspecting users.
Reposted by Dezhe Jin
🚨 We’re hiring: Associate Professor/Professor of Cognitive Neuroscience
🧠 Lead pioneering imaging-informed cognitive neuroscience research
🌍 Open to international applicants | Visa sponsorship available
📍 Permanent, full-time, on-site
🔗 Apply now: www.ucl.ac.uk/work-at-ucl/...
🧠 Lead pioneering imaging-informed cognitive neuroscience research
🌍 Open to international applicants | Visa sponsorship available
📍 Permanent, full-time, on-site
🔗 Apply now: www.ucl.ac.uk/work-at-ucl/...
August 11, 2025 at 3:53 PM
🚨 We’re hiring: Associate Professor/Professor of Cognitive Neuroscience
🧠 Lead pioneering imaging-informed cognitive neuroscience research
🌍 Open to international applicants | Visa sponsorship available
📍 Permanent, full-time, on-site
🔗 Apply now: www.ucl.ac.uk/work-at-ucl/...
🧠 Lead pioneering imaging-informed cognitive neuroscience research
🌍 Open to international applicants | Visa sponsorship available
📍 Permanent, full-time, on-site
🔗 Apply now: www.ucl.ac.uk/work-at-ucl/...
Reposted by Dezhe Jin
Is it possible to go from spikes to rates without averaging?
We show how to exactly map recurrent spiking networks into recurrent rate networks, with the same number of neurons. No temporal or spatial averaging needed!
Presented at Gatsby Neural Dynamics Workshop, London.
We show how to exactly map recurrent spiking networks into recurrent rate networks, with the same number of neurons. No temporal or spatial averaging needed!
Presented at Gatsby Neural Dynamics Workshop, London.
From Spikes To Rates
YouTube video by Gerstner Lab
youtu.be
August 8, 2025 at 3:25 PM
Is it possible to go from spikes to rates without averaging?
We show how to exactly map recurrent spiking networks into recurrent rate networks, with the same number of neurons. No temporal or spatial averaging needed!
Presented at Gatsby Neural Dynamics Workshop, London.
We show how to exactly map recurrent spiking networks into recurrent rate networks, with the same number of neurons. No temporal or spatial averaging needed!
Presented at Gatsby Neural Dynamics Workshop, London.
GPT-5 thinking nailed the problem of numerically solving the radial wave function for a particle in 2D harmonic trap. In contrast, the previous O4-high and Google’s Gemini 2.5 pro all failed. GPT-5 does feel like a big upgrade.
August 8, 2025 at 9:27 PM
GPT-5 thinking nailed the problem of numerically solving the radial wave function for a particle in 2D harmonic trap. In contrast, the previous O4-high and Google’s Gemini 2.5 pro all failed. GPT-5 does feel like a big upgrade.
Reposted by Dezhe Jin
Reposted by Dezhe Jin
Spiking neural networks people, this message is for you!
The annual SNUFA workshop is now open for abstract submission (deadline Sept 26) and (free) registration. This year's speakers include Elisabetta Chicca, Jason Eshraghian, Tomoki Fukai, Chengcheng Huang, and... you?
snufa.net/2025/
🤖🧠🧪
The annual SNUFA workshop is now open for abstract submission (deadline Sept 26) and (free) registration. This year's speakers include Elisabetta Chicca, Jason Eshraghian, Tomoki Fukai, Chengcheng Huang, and... you?
snufa.net/2025/
🤖🧠🧪
SNUFA 2025
Spiking Neural networks as Universal Function Approximators
snufa.net
August 7, 2025 at 11:56 AM
Spiking neural networks people, this message is for you!
The annual SNUFA workshop is now open for abstract submission (deadline Sept 26) and (free) registration. This year's speakers include Elisabetta Chicca, Jason Eshraghian, Tomoki Fukai, Chengcheng Huang, and... you?
snufa.net/2025/
🤖🧠🧪
The annual SNUFA workshop is now open for abstract submission (deadline Sept 26) and (free) registration. This year's speakers include Elisabetta Chicca, Jason Eshraghian, Tomoki Fukai, Chengcheng Huang, and... you?
snufa.net/2025/
🤖🧠🧪
Reposted by Dezhe Jin
Reposted by Dezhe Jin
There is no binding problem because vision—and cortex more broadly—is not modular. Phrenological theories create phrenological artifacts.
Beyond binding: from modular to natural vision
doi.org/10.1016/j.ti...
#neuroscience
Beyond binding: from modular to natural vision
doi.org/10.1016/j.ti...
#neuroscience
Redirecting
doi.org
June 3, 2025 at 7:55 PM
There is no binding problem because vision—and cortex more broadly—is not modular. Phrenological theories create phrenological artifacts.
Beyond binding: from modular to natural vision
doi.org/10.1016/j.ti...
#neuroscience
Beyond binding: from modular to natural vision
doi.org/10.1016/j.ti...
#neuroscience
Reposted by Dezhe Jin
Nature research paper: Observing anyonization of bosons in a quantum gas
https://go.nature.com/3HgWUWU
https://go.nature.com/3HgWUWU
Observing anyonization of bosons in a quantum gas - Nature
Emergent anyonic correlations via spin–charge separation are observed in a one-dimensional strongly interacting quantum gas, enabling the exploration of non-equilibrium anyonic phenomena in a highly controllable setting.
go.nature.com
May 28, 2025 at 4:48 PM
Nature research paper: Observing anyonization of bosons in a quantum gas
https://go.nature.com/3HgWUWU
https://go.nature.com/3HgWUWU
Reposted by Dezhe Jin
Go on parrots
Pet parrots were taught to do video calls, and it was a hit.
The birds not only made friends with each other, and selected who they wanted to call and when, but also picked up new skills from fellow parrots.
Good enrichment option for improving captive parrot welfare.
🧪🦜
dl.acm.org/doi/10.1145/...
The birds not only made friends with each other, and selected who they wanted to call and when, but also picked up new skills from fellow parrots.
Good enrichment option for improving captive parrot welfare.
🧪🦜
dl.acm.org/doi/10.1145/...
May 12, 2025 at 4:40 PM
Go on parrots
Reposted by Dezhe Jin
Very well-articulated piece about how every American wins from federally funded innovative science
RIP, American innovation. It’s been great to be from the most inventive country on earth. My eulogy for what’s being lost as political leaders slash science research budgets. For @postopinions.bsky.social (gift link) wapo.st/4keFurS
Opinion | RIP American innovation
Why destroy the funding that made the United States a leader in technology and invention?
wapo.st
May 12, 2025 at 4:55 PM
Very well-articulated piece about how every American wins from federally funded innovative science
Reposted by Dezhe Jin
Why are grid cells hexagonal? Turns out, it might be the only way to encode 2D paths with neural sequences.
elifesciences.org/reviewed-pre...
elifesciences.org/reviewed-pre...
Spatial periodicity in grid cell firing is explained by a neural sequence code of 2-D trajectories
elifesciences.org
May 8, 2025 at 4:05 PM
Why are grid cells hexagonal? Turns out, it might be the only way to encode 2D paths with neural sequences.
elifesciences.org/reviewed-pre...
elifesciences.org/reviewed-pre...
Reposted by Dezhe Jin
Our paper is out in Nature.
By examining various inputs to the motor cortex during learning, we found that thalamic inputs learn to activate the cortical neurons encoding the movement being learned.
Tour de force by Assaf in collab with Felix and Marcus. Congrats!
www.nature.com/articles/s41...
By examining various inputs to the motor cortex during learning, we found that thalamic inputs learn to activate the cortical neurons encoding the movement being learned.
Tour de force by Assaf in collab with Felix and Marcus. Congrats!
www.nature.com/articles/s41...
Motor learning refines thalamic influence on motor cortex - Nature
Imaging and optogenetics in mice provide insight into the interplay between the primary motor cortex and the motor thalamus during learning, showing that thalamic inputs have a key role in the executi...
www.nature.com
May 8, 2025 at 12:12 AM
Our paper is out in Nature.
By examining various inputs to the motor cortex during learning, we found that thalamic inputs learn to activate the cortical neurons encoding the movement being learned.
Tour de force by Assaf in collab with Felix and Marcus. Congrats!
www.nature.com/articles/s41...
By examining various inputs to the motor cortex during learning, we found that thalamic inputs learn to activate the cortical neurons encoding the movement being learned.
Tour de force by Assaf in collab with Felix and Marcus. Congrats!
www.nature.com/articles/s41...
Reposted by Dezhe Jin
Testing the social complexity hypothesis in monk parakeets: are there relations b/t social networks and vocal diversity?
Surprisingly, hypotheses not supported, but data show birds are learning new contact calls.
🤔 Do we need to redefine "complex"/"diversity"/"info"?
#prattle 💬
#bioacoustics
Surprisingly, hypotheses not supported, but data show birds are learning new contact calls.
🤔 Do we need to redefine "complex"/"diversity"/"info"?
#prattle 💬
#bioacoustics
May 8, 2025 at 4:38 PM
Testing the social complexity hypothesis in monk parakeets: are there relations b/t social networks and vocal diversity?
Surprisingly, hypotheses not supported, but data show birds are learning new contact calls.
🤔 Do we need to redefine "complex"/"diversity"/"info"?
#prattle 💬
#bioacoustics
Surprisingly, hypotheses not supported, but data show birds are learning new contact calls.
🤔 Do we need to redefine "complex"/"diversity"/"info"?
#prattle 💬
#bioacoustics