Masafumi Oizumi

This study compares how different learning paradigms enable deep neural networks to predict human color perception across a vast color space.

This paper proposes a data-driven framework for optimally designing perturbation inputs to improve neural system identification and control.

This paper improves recovery of neural oscillatory modes from partial recordings using delay embedding and perturbation, showing increased mode estimates with deeper delay and perturbation.

The paper uses unsupervised alignment to show that preschoolers and adults share a similar colour experience structure, while toddlers show less alignment, especially when removing late-acquired colour terms.

The paper presents a dataset of single-cell calcium imaging from multiple cortical areas in mice across different physiological states.

The paper maps color similarity structures between individuals to determine if their color experiences align relationally without using color labels.

The paper investigates network structure changes during sleep and anesthesia at single-cell and mesoscales, finding increased modularity at the single-cell level and spatially intermixed modules across states.

This study introduces a method to estimate the controllability Gramian matrix using perturbation inputs and network control theory to analyze brain dynamics, showing that controllable directions better differentiate brain states than overall controllability.

The paper shows that intrinsic neural dynamics in the right temporo-parieto-occipital junction predict upcoming visual gaze targets during low-saliency periods in movie watching.

This paper introduces an unsupervised alignment method using Gromov-Wasserstein Optimal Transport to compare neural representations of natural scenes across individuals and brain areas, finding alignment within visual cortical areas but not across different areas.

This study identifies brain network cores with strong bidirectional interactions using fMRI data and links them to conscious perception and sensorimotor functions.

This paper introduces an unsupervised alignment method using Gromov-Wasserstein optimal transport for neuroscience, showing its ability to reveal structural distinctions in representations across different data domains.

The paper introduces a framework to quantify control costs in brain state transitions using a linear stochastic model, identifying key brain regions involved in controlling mean and covariance of neural activity.

The paper identifies brain subnetworks essential for consciousness by analyzing bidirectional connections in a mouse connectome.

The paper introduces a framework to quantify brain state transition cost using stochastic optimal control theory and applies it to fMRI data to show a link between transition cost and task difficulty.

This paper introduces an algorithm to efficiently identify essential sub-networks (complexes) in brain networks using an information-theoretic approach, demonstrating its application to electrocorticogram data.

The paper introduces FASHIO-2PM, a two-photon microscope that enables fast, aberration-free calcium imaging of large neural networks in awake mice at 7.5 Hz.

The paper proposes a spectral decomposition of integrated information in linear autoregressive processes to analyze frequency-specific causal influences.

The paper tests if integrated information theory applies to conscious experience by analyzing brain recordings and finding that integrated information structures classify visual experiences more accurately than other measures.

This study uses fMRI to show that neural response differentiation, measured by Lempel-Ziv complexity, correlates with stimulus set meaningfulness, with movies showing the highest differentiation.