Shunsuke Sakai

About

I am a 2nd year graduate school student at the University of Fukui, Japan. I aim to democratize high-order artificial intelligence.

Research Interests

Computer Vision
- Visual Representation Learning
- Vision-Language Models
- Visual Anomaly Detection
- Ego-centric Learning
Reinforcement Learning
- World Models
- Unsupervised Reinforcement Learning
- Large Reasoning Models
Generative Modeling
- Diffusion Models
- Flow Matching
- Diffusion Language Models
Spiking Neural Networks
AI Energetics

Publications

LADMIM: Logical Anomaly Detection with Masked Image Modeling in Discrete Latent Space

LADMIM: Logical Anomaly Detection with Masked Image Modeling in Discrete Latent Space

Shunsuke Sakai, Tatsuhito Hasegawa, Makoto Koshino

Under review

ABS HTML PDF CODE

Reconstruction-Free Anomaly Detection with Diffusion Models via Direct Latent Likelihood Evaluation

Reconstruction-Free Anomaly Detection with Diffusion Models via Direct Latent Likelihood Evaluation

Shunsuke Sakai, Tatsuhito Hasegawa

Under review

ABS HTML PDF CODE

Noisy Deep Ensemble: Accelerating Deep Ensemble Learning via Noise Injection

Noisy Deep Ensemble: Accelerating Deep Ensemble Learning via Noise Injection

Shunsuke Sakai, Shunsuke Tsuge, Tatsuhito Hasegawa

ICONIP2024, oral

ABS HTML PDF CODE

Analytical Softmax Temperature Setting from Feature Dimensions for Model- and Domain-Robust Classification

Analytical Softmax Temperature Setting from Feature Dimensions for Model- and Domain-Robust Classification

Tatsuhito Hasegawa, Shunsuke Sakai

Under review

Contrastive Learning-Enhanced Trajectory Matching for Small-Scale Dataset Distillation

Contrastive Learning-Enhanced Trajectory Matching for Small-Scale Dataset Distillation

Wenmin Li, Shunsuke Sakai, Tatsuhito Hasegawa

Under review