Lightnews — Scholar-powered news

LightNews

Yash Shah

@ynshah.bsky.social

200 followers 1.3K following 12 posts

PhD student at Stanford. Self-proclaimed computational neuroscientist and humanist. Incomplete bio at https://ynshah3.github.io/.

ynshah3.github.io

Posts Media Videos Starter Packs

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 6d

Representations in the hippocampal-entorhinal system emerge from learning sensory predictions https://www.biorxiv.org/content/10.1101/2025.10.03.680189v1

Reposted by Yash Shah

Robert Hawkins @rdhawkins.bsky.social · 24d

Very exciting preprint from Dan Yamins' NeuroAI lab, proposing Probabilistic Structure Integration (PSI), a way to bootstrap from pixels to higher-level visual abstractions through a kind of visual prompting. One of the deepest and most original ideas I've read in a while.

arxiv.org/abs/2509.09737

World Modeling with Probabilistic Structure Integration

We present Probabilistic Structure Integration (PSI), a system for learning richly controllable and flexibly promptable world models from data. PSI consists of a three-step cycle. The first step, Prob...

Reposted by Yash Shah

Martin Schrimpf @mschrimpf.bsky.social · 7d

I've been arguing that #NeuroAI should model the brain in health *and* in disease -- very excited to share a first step from Melika Honarmand: inducing dyslexia in vision-language-models via targeted perturbations of visual-word-form units (analogous to human VWFA) 🧠🤖🧪 arxiv.org/abs/2509.24597

Reposted by Yash Shah

arXiv q-bio.NC Neurons and Cognition @qbionc-bot.bsky.social · 10d

Haider Al-Tahan, Mayukh Deb, Jenelle Feather, N. Apurva Ratan Murty: End-to-end Topographic Auditory Models Replicate Signatures of Human Auditory Cortex https://arxiv.org/abs/2509.24039 https://arxiv.org/pdf/2509.24039 https://arxiv.org/html/2509.24039

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 10d

Metabolic organization of macaque visual cortex reflects retinotopic eccentricity and category selectivity https://www.biorxiv.org/content/10.1101/2025.09.27.678945v1

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · 17d

Functional organization of the human visual system at birth and across late gestation https://www.biorxiv.org/content/10.1101/2025.09.22.677834v1

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Aug 4

Unfolding spatiotemporal representations of 3D visual perception in the human brain https://www.biorxiv.org/content/10.1101/2025.08.03.668371v1

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Jul 27

Visual Word Form Area demonstrates individual and task-agnostic consistency but inter-individual variability https://www.biorxiv.org/content/10.1101/2025.07.23.666206v1

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Jul 26

Many-Two-One: Diverse Representations Across Visual Pathways Emerge from A Single Objective https://www.biorxiv.org/content/10.1101/2025.07.22.664908v1

Yash Shah @ynshah.bsky.social · Jul 13

And of course because this is my first ever post I forgot to include hashtags! #ICML2025

Yash Shah @ynshah.bsky.social · Jul 13

Check out the paper if interested and come talk to me during the poster session (July 17, Thursday at 4:30pm) if in Vancouver! icml.cc/virtual/2025.... [11/n]

ICML Poster Confounder-Free Continual Learning via Recursive Feature NormalizationICML 2025

Yash Shah @ynshah.bsky.social · Jul 13

Finally, R-MDN, because it operates on the level of individual examples, can be integrated in both convolutional neural networks and vision transformers—which was one of the significant limitations of the MDN algorithm. [10/n]

Yash Shah @ynshah.bsky.social · Jul 13

And R-MDN makes equitable predictions across population groups, such as across both boys and girls when performing sex classification on the ABCD (Casey et al., 2008) dataset in the presence of pubertal development scores as the confounder. [9/n]

Yash Shah @ynshah.bsky.social · Jul 13

R-MDN can also remove the influence from multiple confounding variables, as seen when testing on the ADNI (Mueller et al., 2005) dataset. [8/n]

Yash Shah @ynshah.bsky.social · Jul 13

Since R-MDN is a normalization layer, it can be tacked on to various already-proposed model architectures. [7/n]

Yash Shah @ynshah.bsky.social · Jul 13

R-MDN effectively removes confounder influence from learned DNN features, as rigorously verified in both synthetically controlled environments and real-world datasets. [6/n]

Yash Shah @ynshah.bsky.social · Jul 13

We propose Recursive Metadata Normalization (R-MDN), a normalization layer that leverages the statistical recursive least squares algorithm to iteratively update its internal parameters based on previously computed values whenever new data are received. [5/n]

Yash Shah @ynshah.bsky.social · Jul 13

However, within continual learning, data becomes available sequentially, often over the span of several years, as in longitudinal studies. [4/n]

Yash Shah @ynshah.bsky.social · Jul 13

Prior work such as BR-Net (Adeli et al., 2020), MDN (Lu et al., 2021), and P-MDN (Vento et al., 2022) proposed to learn confounder-invariant representations in DNNs work within a static learning setting and assume that the algorithm has access to all data at the outset of training. [3/n]

Yash Shah @ynshah.bsky.social · Jul 13

Confounders are variables that influence both the outcome (i.e., the output) and the exposure (i.e., the input) in a study, causing spurious associations. [2/n]

Yash Shah @ynshah.bsky.social · Jul 13

I am excited to share that my work on "Confounder-Free Continual Learning via Recursive Feature Normalization" has been accepted at ICML 2025! Very grateful to @camgonza.bsky.social , @mhabbasi.bsky.social , @qingyuz.bsky.social, Kilian Pohl, and @eadeli.bsky.social for always supporting me. [1/n]

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Jul 12

Retinal waves reveal axial biases in modular patterns of cortical activity that predict future orientation preferences https://www.biorxiv.org/content/10.1101/2025.07.09.663735v1

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · Jun 3

Visual processing of manipulable objects in the ventral stream is modulated by inputs from parietal action systems https://www.biorxiv.org/content/10.1101/2025.06.01.657280v1

Reposted by Yash Shah

Hannes Mehrer @hannesmehrer.bsky.social · May 27

Announcement: Workshop at #CCN2025
🧠 Modeling the Physical Brain: Spatial Organization & Biophysical Constraints
🗓️ Monday, Aug 11 | 🕦 11:30–18:00 CET | 📍 Room A2.07
🔗 Register: tinyurl.com/CCN-physical...
#NeuroAI @cogcompneuro.bsky.social

Reposted by Yash Shah

bioRxiv Neuroscience @biorxiv-neursci.bsky.social · May 23

A two-dimensional space of linguistic representations shared across individuals https://www.biorxiv.org/content/10.1101/2025.05.21.655330v1