Salvato in:
Dettagli Bibliografici
Autori principali: Mao, Ziming, Xu, Jia, Zheng, Zeqi, Zheng, Haofang, Sheng, Dabing, Jin, Yaochu, Yang, Guoyuan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2506.11123
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912427296161792
author Mao, Ziming
Xu, Jia
Zheng, Zeqi
Zheng, Haofang
Sheng, Dabing
Jin, Yaochu
Yang, Guoyuan
author_facet Mao, Ziming
Xu, Jia
Zheng, Zeqi
Zheng, Haofang
Sheng, Dabing
Jin, Yaochu
Yang, Guoyuan
contents We present SAE-BrainMap, a novel framework that directly aligns deep learning visual model representations with voxel-level fMRI responses using sparse autoencoders (SAEs). First, we train layer-wise SAEs on model activations and compute the correlations between SAE unit activations and cortical fMRI signals elicited by the same natural image stimuli with cosine similarity, revealing strong activation correspondence (maximum similarity up to 0.76). Depending on this alignment, we construct a voxel dictionary by optimally assigning the most similar SAE feature to each voxel, demonstrating that SAE units preserve the functional structure of predefined regions of interest (ROIs) and exhibit ROI-consistent selectivity. Finally, we establish fine-grained hierarchical mapping between model layers and the human ventral visual pathway, also by projecting voxel dictionary activations onto individual cortical surfaces, we visualize the dynamic transformation of the visual information in deep learning models. It is found that ViT-B/16$_{CLIP}$ tends to utilize low-level information to generate high-level semantic information in the early layers and reconstructs the low-dimension information later. Our results establish a direct, downstream-task-free bridge between deep neural networks and human visual cortex, offering new insights into model interpretability.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11123
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Sparse Autoencoders Bridge The Deep Learning Model and The Brain
Mao, Ziming
Xu, Jia
Zheng, Zeqi
Zheng, Haofang
Sheng, Dabing
Jin, Yaochu
Yang, Guoyuan
Neurons and Cognition
Computer Vision and Pattern Recognition
We present SAE-BrainMap, a novel framework that directly aligns deep learning visual model representations with voxel-level fMRI responses using sparse autoencoders (SAEs). First, we train layer-wise SAEs on model activations and compute the correlations between SAE unit activations and cortical fMRI signals elicited by the same natural image stimuli with cosine similarity, revealing strong activation correspondence (maximum similarity up to 0.76). Depending on this alignment, we construct a voxel dictionary by optimally assigning the most similar SAE feature to each voxel, demonstrating that SAE units preserve the functional structure of predefined regions of interest (ROIs) and exhibit ROI-consistent selectivity. Finally, we establish fine-grained hierarchical mapping between model layers and the human ventral visual pathway, also by projecting voxel dictionary activations onto individual cortical surfaces, we visualize the dynamic transformation of the visual information in deep learning models. It is found that ViT-B/16$_{CLIP}$ tends to utilize low-level information to generate high-level semantic information in the early layers and reconstructs the low-dimension information later. Our results establish a direct, downstream-task-free bridge between deep neural networks and human visual cortex, offering new insights into model interpretability.
title Sparse Autoencoders Bridge The Deep Learning Model and The Brain
topic Neurons and Cognition
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.11123