Alisa Leshchenko 🦕
@borromeanlink.bsky.social
CompNeuro @Columbia Zuckerman Institute, Fusi Lab | Cognitive maps, abstraction, compositionality in a neural substrate | #NeuroAI
Reposted by Alisa Leshchenko 🦕
Vase with Irises - 1890
https://botfrens.com/collections/46/contents/15543
https://botfrens.com/collections/46/contents/15543
October 31, 2025 at 6:15 PM
Vase with Irises - 1890
https://botfrens.com/collections/46/contents/15543
https://botfrens.com/collections/46/contents/15543
Reposted by Alisa Leshchenko 🦕
LLMs are trained to compress data by mapping sequences to high-dim representations!
How does the complexity of this mapping change across LLM training? How does it relate to the model’s capabilities? 🤔
Announcing our #NeurIPS2025 📄 that dives into this.
🧵below
#AIResearch #MachineLearning #LLM
How does the complexity of this mapping change across LLM training? How does it relate to the model’s capabilities? 🤔
Announcing our #NeurIPS2025 📄 that dives into this.
🧵below
#AIResearch #MachineLearning #LLM
October 31, 2025 at 4:19 PM
LLMs are trained to compress data by mapping sequences to high-dim representations!
How does the complexity of this mapping change across LLM training? How does it relate to the model’s capabilities? 🤔
Announcing our #NeurIPS2025 📄 that dives into this.
🧵below
#AIResearch #MachineLearning #LLM
How does the complexity of this mapping change across LLM training? How does it relate to the model’s capabilities? 🤔
Announcing our #NeurIPS2025 📄 that dives into this.
🧵below
#AIResearch #MachineLearning #LLM
Reposted by Alisa Leshchenko 🦕
Excited to share a new preprint w/ @annaschapiro.bsky.social! Why are there gradients of plasticity and sparsity along the neocortex–hippocampus hierarchy? We show that brain-like organization of these properties emerges in ANNs that meta-learn layer-wise plasticity and sparsity. bit.ly/4kB1yg5
A gradient of complementary learning systems emerges through meta-learning
Long-term learning and memory in the primate brain rely on a series of hierarchically organized subsystems extending from early sensory neocortical areas to the hippocampus. The components differ in t...
bit.ly
July 16, 2025 at 4:15 PM
Excited to share a new preprint w/ @annaschapiro.bsky.social! Why are there gradients of plasticity and sparsity along the neocortex–hippocampus hierarchy? We show that brain-like organization of these properties emerges in ANNs that meta-learn layer-wise plasticity and sparsity. bit.ly/4kB1yg5
Reposted by Alisa Leshchenko 🦕
Connectome suggests brain’s synaptic weights follow heavy-tailed distributions, yet most analyses of RNNs assume Gaussian connectivity.
🧵⬇️ Our @alleninstitute.org #NeurIPS2025 paper shows heavy-tailed weights can strongly affect dynamics, trade off robustness + attractor dimension.
🧵⬇️ Our @alleninstitute.org #NeurIPS2025 paper shows heavy-tailed weights can strongly affect dynamics, trade off robustness + attractor dimension.
October 30, 2025 at 2:54 PM
Connectome suggests brain’s synaptic weights follow heavy-tailed distributions, yet most analyses of RNNs assume Gaussian connectivity.
🧵⬇️ Our @alleninstitute.org #NeurIPS2025 paper shows heavy-tailed weights can strongly affect dynamics, trade off robustness + attractor dimension.
🧵⬇️ Our @alleninstitute.org #NeurIPS2025 paper shows heavy-tailed weights can strongly affect dynamics, trade off robustness + attractor dimension.
Reposted by Alisa Leshchenko 🦕
**Discovering network dynamics**
One more on estimating dynamics of complex systems, this time with symbolic regression
doi.org/10.1038/s435...
One more on estimating dynamics of complex systems, this time with symbolic regression
doi.org/10.1038/s435...
October 30, 2025 at 4:57 PM
**Discovering network dynamics**
One more on estimating dynamics of complex systems, this time with symbolic regression
doi.org/10.1038/s435...
One more on estimating dynamics of complex systems, this time with symbolic regression
doi.org/10.1038/s435...
Reposted by Alisa Leshchenko 🦕
Total Lunar Eclipse - 26 May 2021 - From Ángel López-Sánchez - https://flic.kr/p/2m1PLyZ
October 16, 2025 at 4:00 AM
Total Lunar Eclipse - 26 May 2021 - From Ángel López-Sánchez - https://flic.kr/p/2m1PLyZ
Jorge Luis Borges
October 16, 2025 at 4:19 AM
Jorge Luis Borges
Reposted by Alisa Leshchenko 🦕
Happy birthday Kay Sage. The artist & poet, famous for her surrealist paintings, often referencing architecture & the built environment, was born today in 1898. #surrealism
June 25, 2025 at 5:13 PM
Happy birthday Kay Sage. The artist & poet, famous for her surrealist paintings, often referencing architecture & the built environment, was born today in 1898. #surrealism
Reposted by Alisa Leshchenko 🦕
The Eye of Silence https://www.wikiart.org/en/max-ernst/the-eye-of-silence-1943
October 9, 2025 at 3:31 PM
The Eye of Silence https://www.wikiart.org/en/max-ernst/the-eye-of-silence-1943
Reposted by Alisa Leshchenko 🦕
What do we talk about when we talk about "readout"?
I argued that our overly specialized, modular approach to studying the brain has given us a simplistic view of readout.
🧠📈
I argued that our overly specialized, modular approach to studying the brain has given us a simplistic view of readout.
🧠📈
Figuring out how the brain uses information from visual neurons may require new tools, writes @neurograce.bsky.social. Hear from 10 experts in the field.
#neuroskyence
www.thetransmitter.org/the-big-pict...
#neuroskyence
www.thetransmitter.org/the-big-pict...
Connecting neural activity, perception in the visual system
Figuring out how the brain uses information from visual neurons may require new tools. I asked nine experts to weigh in.
www.thetransmitter.org
October 13, 2025 at 3:15 PM
What do we talk about when we talk about "readout"?
I argued that our overly specialized, modular approach to studying the brain has given us a simplistic view of readout.
🧠📈
I argued that our overly specialized, modular approach to studying the brain has given us a simplistic view of readout.
🧠📈
Reposted by Alisa Leshchenko 🦕
Prominence 10-26-10 - From Jason Major (jpmajor.bsky.social) - https://flic.kr/p/8NkYgk
October 14, 2025 at 9:00 AM
Prominence 10-26-10 - From Jason Major (jpmajor.bsky.social) - https://flic.kr/p/8NkYgk
Reposted by Alisa Leshchenko 🦕
ESA ROSETTA 14 July 2015 - From 2di7 & titanio44 - https://flic.kr/p/vpqn2z
October 15, 2025 at 9:00 AM
ESA ROSETTA 14 July 2015 - From 2di7 & titanio44 - https://flic.kr/p/vpqn2z
Reposted by Alisa Leshchenko 🦕
Mimas, Epimetheus and Rings - From Gordan Ugarković (ugordan.bsky.social) - https://flic.kr/p/5oPunj
October 15, 2025 at 1:00 PM
Mimas, Epimetheus and Rings - From Gordan Ugarković (ugordan.bsky.social) - https://flic.kr/p/5oPunj
Reposted by Alisa Leshchenko 🦕
🧠🚨 How does the hippocampus transform the visual similarity space to resolve memory interference?
In this new preprint, we found that the hippocampus sequentially inverts the behaviorally relevant dimensions of similarity 🧵
www.biorxiv.org/content/10.1...
In this new preprint, we found that the hippocampus sequentially inverts the behaviorally relevant dimensions of similarity 🧵
www.biorxiv.org/content/10.1...
Hippocampal transformations occur along dimensions of memory interference
The role of the hippocampus in resolving memory interference has been greatly elucidated by considering the relationship between the similarity of visual stimuli (input) and corresponding similarity o...
www.biorxiv.org
October 14, 2025 at 4:48 PM
🧠🚨 How does the hippocampus transform the visual similarity space to resolve memory interference?
In this new preprint, we found that the hippocampus sequentially inverts the behaviorally relevant dimensions of similarity 🧵
www.biorxiv.org/content/10.1...
In this new preprint, we found that the hippocampus sequentially inverts the behaviorally relevant dimensions of similarity 🧵
www.biorxiv.org/content/10.1...
Reposted by Alisa Leshchenko 🦕
Replace Variational Autoencoder (VAE) with pretrained representation encoders (e.g., DINO, SigLIP, MAE) paired with trained decoders, which they terms as Representation Autoencoders (RAE).
October 15, 2025 at 3:49 AM
Replace Variational Autoencoder (VAE) with pretrained representation encoders (e.g., DINO, SigLIP, MAE) paired with trained decoders, which they terms as Representation Autoencoders (RAE).
Reposted by Alisa Leshchenko 🦕
Preprint Alert 🚀
Can we simultaneously learn transformation-invariant and transformation-equivariant representations with self-supervised learning?
TL;DR Yes! This is possible via simple predictive learning & architectural inductive biases – without extra loss terms and predictors!
🧵 (1/10)
Can we simultaneously learn transformation-invariant and transformation-equivariant representations with self-supervised learning?
TL;DR Yes! This is possible via simple predictive learning & architectural inductive biases – without extra loss terms and predictors!
🧵 (1/10)
May 14, 2025 at 12:53 PM
Preprint Alert 🚀
Can we simultaneously learn transformation-invariant and transformation-equivariant representations with self-supervised learning?
TL;DR Yes! This is possible via simple predictive learning & architectural inductive biases – without extra loss terms and predictors!
🧵 (1/10)
Can we simultaneously learn transformation-invariant and transformation-equivariant representations with self-supervised learning?
TL;DR Yes! This is possible via simple predictive learning & architectural inductive biases – without extra loss terms and predictors!
🧵 (1/10)
Reposted by Alisa Leshchenko 🦕
So excited to see this preprint released from the lab into the wild.
Charlotte has developed a theory for how learning curriculum influences learning generalization.
Our theory makes straightforward neural predictions that can be tested in future experiments. (1/4)
🧠🤖 🧠📈 #MLSky
Charlotte has developed a theory for how learning curriculum influences learning generalization.
Our theory makes straightforward neural predictions that can be tested in future experiments. (1/4)
🧠🤖 🧠📈 #MLSky
🚨 New preprint alert!
🧠🤖
We propose a theory of how learning curriculum affects generalization through neural population dimensionality. Learning curriculum is a determining factor of neural dimensionality - where you start from determines where you end up.
🧠📈
A 🧵:
tinyurl.com/yr8tawj3
🧠🤖
We propose a theory of how learning curriculum affects generalization through neural population dimensionality. Learning curriculum is a determining factor of neural dimensionality - where you start from determines where you end up.
🧠📈
A 🧵:
tinyurl.com/yr8tawj3
The curriculum effect in visual learning: the role of readout dimensionality
Generalization of visual perceptual learning (VPL) to unseen conditions varies across tasks. Previous work suggests that training curriculum may be integral to generalization, yet a theoretical explan...
tinyurl.com
September 30, 2025 at 2:35 PM
So excited to see this preprint released from the lab into the wild.
Charlotte has developed a theory for how learning curriculum influences learning generalization.
Our theory makes straightforward neural predictions that can be tested in future experiments. (1/4)
🧠🤖 🧠📈 #MLSky
Charlotte has developed a theory for how learning curriculum influences learning generalization.
Our theory makes straightforward neural predictions that can be tested in future experiments. (1/4)
🧠🤖 🧠📈 #MLSky
Reposted by Alisa Leshchenko 🦕
Tomorrow the next meeting of MIT #Consciousness Club. This is a Zoom link 🔗⬇️
The next session of the MIT Consciousness Club is on Thursday 16, 12pm-1:30pm. Rachel Denison will present “Attentional Distortions of Subjective Perception”. More information here: sites.google.com/view/mit-con....
MIT Consciousness Club
The MIT Consciousness Club aims to foster interdisciplinary research on consciousness at MIT and in the broader Boston area by organizing a monthly event featuring an expert talk on consciousness foll...
sites.google.com
October 15, 2025 at 9:38 PM
Tomorrow the next meeting of MIT #Consciousness Club. This is a Zoom link 🔗⬇️
Reposted by Alisa Leshchenko 🦕
Max Ernst Showing a Young Girl the Head of his Father https://www.wikiart.org/en/max-ernst/max-ernst-showing-a-young-girl-the-head-of-his-father
October 13, 2025 at 11:54 AM
Max Ernst Showing a Young Girl the Head of his Father https://www.wikiart.org/en/max-ernst/max-ernst-showing-a-young-girl-the-head-of-his-father
Reposted by Alisa Leshchenko 🦕
Cielo di piombo, ispettore Callaghan...
October 5, 2025 at 11:44 AM
Cielo di piombo, ispettore Callaghan...
Reposted by Alisa Leshchenko 🦕
Reposted by Alisa Leshchenko 🦕
The domain of Arnheim (1962)
by René Magritte
by René Magritte
October 13, 2025 at 9:33 AM
The domain of Arnheim (1962)
by René Magritte
by René Magritte
Reposted by Alisa Leshchenko 🦕
Reposted by Alisa Leshchenko 🦕
Reposted by Alisa Leshchenko 🦕
Interesting paper!
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Thalamus enables active dendritic coupling of inputs arriving at different cortical layers - Nature Communications
How the brain combines sensory with non-sensory information is unknown. Here, the authors find that sensory input from the thalamus to the apical dendrites of the main cortical output neurons enables ...
www.nature.com
October 15, 2025 at 7:49 PM
Interesting paper!
www.nature.com/articles/s41...
www.nature.com/articles/s41...