Projects

My research focuses on understanding human vision and cognition, and on improving the quantity and quality of AI data.

Panasonic-LLM-100b

Development of Japan's largest scale (100 billion parameters) Japanese LLM with Stockmark Inc.

Diffusion-KTO

Aligning text-to-image diffusion models by formulating the alignment objective as the maximization of expected human utility

Wild2Avatar

A method to render high fidelity human avatars from in-the-wild monocular videos behind occlusions

HIPIE

Hierarchical Open-vocabulary Universal Image Segmentation

Invisible-to-Visible

Privacy-Aware Human Segmentation using Airborne Ultrasound via Collaborative Learning Probabilistic U-Net

Home Action Genome

A large-scale multi-view video dataset of daily activities at home

CFLOW-AD

Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flows

Contrastive Neural Processes

A new method for self-supervised learning that does not require augmentation engineering

AutoDO

Automatically determine data augumentation parameters for a small number of data

Risk Prediction

Training algorithm that enable prediction for ambiguous recognition targets

Medical Image Retrieval

A technology to search for similar cases by reproducing the points that doctors focus on during diagnosis

Brain Machine Interface

An estimate technology using Brain Wave Patterns for Acceptable Maximum Sound Volume of Hearing Aids

Multiple View Geometry

Multiple view geometry between sensors that project to different dimensions