Fei Gao 高飞
@feigao.bsky.social
RA at NCCR Evolving Language｜ETH Zurich | NYU CLaME
I work on AI, Language, Music, and Neuroscience.
I work on AI, Language, Music, and Neuroscience.
Reposted by Fei Gao 高飞
cool paper about compression-meaning trade-off!
They show how LLMs optimize for statistical compression, while humans for adaptive richness (preserve flexibility and context). Also that smaller models outperformed larger models at forming conceptual categories
#neuroAI
arxiv.org/abs/2505.17117
They show how LLMs optimize for statistical compression, while humans for adaptive richness (preserve flexibility and context). Also that smaller models outperformed larger models at forming conceptual categories
#neuroAI
arxiv.org/abs/2505.17117
From Tokens to Thoughts: How LLMs and Humans Trade Compression for Meaning
Humans organize knowledge into compact categories through semantic compression by mapping diverse instances to abstract representations while preserving meaning (e.g., robin and blue jay are both bird...
arxiv.org
June 1, 2025 at 12:02 PM
cool paper about compression-meaning trade-off!
They show how LLMs optimize for statistical compression, while humans for adaptive richness (preserve flexibility and context). Also that smaller models outperformed larger models at forming conceptual categories
#neuroAI
arxiv.org/abs/2505.17117
They show how LLMs optimize for statistical compression, while humans for adaptive richness (preserve flexibility and context). Also that smaller models outperformed larger models at forming conceptual categories
#neuroAI
arxiv.org/abs/2505.17117
Reposted by Fei Gao 高飞
High-Dimensional Dynamics in Low-Dimensional Networks.
New preprint with a former undergrad, Yue Wan.
I'm not totally sure how to talk about these results. They're counterintuitive on the surface, seem somewhat obvious in hindsight, but then there's more to them when you dig deeper.
New preprint with a former undergrad, Yue Wan.
I'm not totally sure how to talk about these results. They're counterintuitive on the surface, seem somewhat obvious in hindsight, but then there's more to them when you dig deeper.
April 21, 2025 at 5:00 PM
High-Dimensional Dynamics in Low-Dimensional Networks.
New preprint with a former undergrad, Yue Wan.
I'm not totally sure how to talk about these results. They're counterintuitive on the surface, seem somewhat obvious in hindsight, but then there's more to them when you dig deeper.
New preprint with a former undergrad, Yue Wan.
I'm not totally sure how to talk about these results. They're counterintuitive on the surface, seem somewhat obvious in hindsight, but then there's more to them when you dig deeper.
Reposted by Fei Gao 高飞
I'm hiring! 2 x PhDs and 1 x PostDoc, to work with me at Uni Zurich on a project investigating how semantic patterns differ by geographic region.
Linguistic typology, data science, spatial analysis, phylogenetic methods, potential fieldwork... REPOST 🙏
www.comparativelinguistics.uzh.ch/en/jobs.html
Linguistic typology, data science, spatial analysis, phylogenetic methods, potential fieldwork... REPOST 🙏
www.comparativelinguistics.uzh.ch/en/jobs.html
Open Positions
www.comparativelinguistics.uzh.ch
April 7, 2025 at 10:01 AM
I'm hiring! 2 x PhDs and 1 x PostDoc, to work with me at Uni Zurich on a project investigating how semantic patterns differ by geographic region.
Linguistic typology, data science, spatial analysis, phylogenetic methods, potential fieldwork... REPOST 🙏
www.comparativelinguistics.uzh.ch/en/jobs.html
Linguistic typology, data science, spatial analysis, phylogenetic methods, potential fieldwork... REPOST 🙏
www.comparativelinguistics.uzh.ch/en/jobs.html
Reposted by Fei Gao 高飞
🚨 New Preprint!!
LLMs trained on next-word prediction (NWP) show high alignment with brain recordings. But what drives this alignment—linguistic structure or world knowledge? And how does this alignment evolve during training? Our new paper explores these questions. 👇🧵
LLMs trained on next-word prediction (NWP) show high alignment with brain recordings. But what drives this alignment—linguistic structure or world knowledge? And how does this alignment evolve during training? Our new paper explores these questions. 👇🧵
March 5, 2025 at 3:58 PM
🚨 New Preprint!!
LLMs trained on next-word prediction (NWP) show high alignment with brain recordings. But what drives this alignment—linguistic structure or world knowledge? And how does this alignment evolve during training? Our new paper explores these questions. 👇🧵
LLMs trained on next-word prediction (NWP) show high alignment with brain recordings. But what drives this alignment—linguistic structure or world knowledge? And how does this alignment evolve during training? Our new paper explores these questions. 👇🧵
Reposted by Fei Gao 高飞
As we try to build artificial agents that can act independently in the real world … adapting their goals as needed, we are discovering that agency is far more complex than it appears, writes @tonyzador.bsky.social in our NeuroAI series.
#neuroskyence
www.thetransmitter.org/neuroai/neur...
#neuroskyence
www.thetransmitter.org/neuroai/neur...
NeuroAI and the hidden complexity of agency
As we attempt to build autonomous AI systems, we’re discovering that a capability we take for granted in animals may be more complex than we imagined.
www.thetransmitter.org
March 29, 2025 at 4:04 PM
As we try to build artificial agents that can act independently in the real world … adapting their goals as needed, we are discovering that agency is far more complex than it appears, writes @tonyzador.bsky.social in our NeuroAI series.
#neuroskyence
www.thetransmitter.org/neuroai/neur...
#neuroskyence
www.thetransmitter.org/neuroai/neur...
Reposted by Fei Gao 高飞
Eva Dyer and I wrote an opinion piece for @thetransmitter.bsky.social on why neuroscience needs to embrace complexity and accept the "bitter lesson" by using a data-driven regime at scale.
With commentary from several wonderful researchers!
🧠📈 #NeuroAI 🧪
With commentary from several wonderful researchers!
🧠📈 #NeuroAI 🧪
How can we make progress in developing a general model of neural computation rather than a series of disjointed models tied to specific experimental circumstances, ask Eva Dyer and @tyrellturing.bsky.social in the latest entry in our NeuroAI series.
www.thetransmitter.org/neuroai/acce...
www.thetransmitter.org/neuroai/acce...
Accepting “bitter lesson” and embracing brain’s complexity
To gain insight into complex neural data, we must move toward a data-driven regime, employing large models trained on vast amounts of data. Experts weigh in.
www.thetransmitter.org
March 26, 2025 at 5:20 PM
Eva Dyer and I wrote an opinion piece for @thetransmitter.bsky.social on why neuroscience needs to embrace complexity and accept the "bitter lesson" by using a data-driven regime at scale.
With commentary from several wonderful researchers!
🧠📈 #NeuroAI 🧪
With commentary from several wonderful researchers!
🧠📈 #NeuroAI 🧪
Reposted by Fei Gao 高飞
🚨 New paper on ArXiv:
“Uncertainty quantification and posterior sampling for network reconstruction”
TL;DR; We present an efficient method to sample the entire ensemble of possible network reconstructions that are compatible with an indirect observation, e.g. a dynamics.
Short thread: 1/N
“Uncertainty quantification and posterior sampling for network reconstruction”
TL;DR; We present an efficient method to sample the entire ensemble of possible network reconstructions that are compatible with an indirect observation, e.g. a dynamics.
Short thread: 1/N
March 12, 2025 at 11:57 AM
🚨 New paper on ArXiv:
“Uncertainty quantification and posterior sampling for network reconstruction”
TL;DR; We present an efficient method to sample the entire ensemble of possible network reconstructions that are compatible with an indirect observation, e.g. a dynamics.
Short thread: 1/N
“Uncertainty quantification and posterior sampling for network reconstruction”
TL;DR; We present an efficient method to sample the entire ensemble of possible network reconstructions that are compatible with an indirect observation, e.g. a dynamics.
Short thread: 1/N
Reposted by Fei Gao 高飞
Decoding the rhythmic representation and communication of visual contents
www.cell.com/trends/neuro...
#neuroscience
www.cell.com/trends/neuro...
#neuroscience
Decoding the rhythmic representation and communication of visual contents
Rhythmic neural activity is considered essential for adaptively modulating responses
in the visual system. In this opinion article we posit that visual brain rhythms also
serve a key function in the r...
www.cell.com
March 12, 2025 at 7:23 PM
Decoding the rhythmic representation and communication of visual contents
www.cell.com/trends/neuro...
#neuroscience
www.cell.com/trends/neuro...
#neuroscience
Reposted by Fei Gao 高飞
We make about 3-4 fast eye movements a second, yet our world appears stable. How is this possible? In a preprint led by @lucakaemmer.bsky.social we test the intriguing idea that anticipatory signals in the fovea may explain visual stability.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
a woman in a white shirt is standing in front of a refrigerator in a store .
ALT: a woman in a white shirt is standing in front of a refrigerator in a store .
media.tenor.com
March 12, 2025 at 10:18 PM
We make about 3-4 fast eye movements a second, yet our world appears stable. How is this possible? In a preprint led by @lucakaemmer.bsky.social we test the intriguing idea that anticipatory signals in the fovea may explain visual stability.
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
Reposted by Fei Gao 高飞
I am incredibly proud to share my first, first-author paper as a postdoc with @benhayden.bsky.social . How does the human hippocampus, known for encoding concepts, represent the meanings of words while listening to narrative speech?
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...
February 27, 2025 at 8:17 PM
I am incredibly proud to share my first, first-author paper as a postdoc with @benhayden.bsky.social . How does the human hippocampus, known for encoding concepts, represent the meanings of words while listening to narrative speech?
www.biorxiv.org/content/10.1...
www.biorxiv.org/content/10.1...