flavie al

0 followers · 625 views

on the atlas — 15

highlights — 11

Our approach was implemented as follows: For the Generic Object Decoding (GOD) dataset (Horikawa and Kamitani [15]), which includes images with specific classes akin to those in ImageNet, we used a methodology involving the linear regression of latent image representations from fMRI data, followed by the use of a k-nearest neighbors (kNN) algorithm for the purposes of image retrieval and classification. Following the classification phase, we expanded our examination to include the generation of images and the assessment of these images by human evaluators
Retrieving and reconstructing conceptually similar images from fMRI with latent diffusion models and a neuro-inspired brain decoding model - IOPscience
In this paper, we focus on context, i.e. the semantic content of presented stimuli, with the aim of reconstructing images that resemble the original ones and can elicit the same fMRI activity.
Retrieving and reconstructing conceptually similar images from fMRI with latent diffusion models and a neuro-inspired brain decoding model - IOPscience
Our work is closely related due to the utilization of the same GOD dataset and latent diffusion models for image reconstruction;
Retrieving and reconstructing conceptually similar images from fMRI with latent diffusion models and a neuro-inspired brain decoding model - IOPscience
Several previous studies have used machine learning classifiers to decode the contents of mental imagery. In a pioneering work from 2009, Harrison and Tong (2009) demonstrated that a classifier trained to predict the orientations of seen (i.e., observed) gratings from fMRI signals could also predict the orientations of gratings remembered in the mind during a working memory task. While this initial study focused on the contents of working memory rather than mental imagery, the same classification approach has subsequently been used to decode the categories of imagined objects and scenes (Alber…
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
A few studies have incorporated Bayesian estimation into the process of visual image reconstruction and demonstrated notable improvements in reconstruction quality.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
By applying the proposed framework to the dataset from Shen et al. (2019), we demonstrate that our framework can reconstruct seen images only using high-level visual information and externalize mental imagery.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
As an alternative approach, we used the stochastic gradient Langevin dynamics (SGLD) algorithm (Welling & Teh, 2011) to sample images from the posterior distribution. Our results demonstrated that seen and imagined images can be successfully reconstructed from brain activity, supporting the effectiveness of the SGLD algorithm in the field of neural decoding.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
we first extended this previous method to a Bayesian estimation framework and then introduced the assistance of semantic information.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
According to other neuroimaging studies, high-level or semantic information (representation) is thought to be recruited more strongly in the brain during mental imagery than low-level visual information. Although low-level visual features of imagined images (e.g., Gabor-wavelet features) can be decoded to a certain extent (Albers et al., 2013; Harrison & Tong, 2009; Naselaris et al., 2015; Xing et al., 2013), high-level visual features are more helpful in identifying imagined objects from brain activity (Horikawa & Kamitani, 2017). Furthermore, categories of imagined objects can be better pred…
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
Previous studies have succeeded in reconstructing images seen by humans from their brain activity; however, externalizing mental imagery remains a challenge.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect
These results suggest that our framework would provide a unique tool for directly investigating the subjective contents of the brain such as illusions, hallucinations, and dreams.
Mental image reconstruction from human brain activity: Neural decoding of mental imagery via deep neural network-based Bayesian estimation - ScienceDirect