Machine Learning | KAIST MatriX

Tag: Machine Learning

Research Highlight

Deep Generative Diffusion Models

AAILab, KAIST has developed various deep generative model structures and inference methods to synthesize realistic images through diffusion processes. ...read more

Deep Generative Models Diffusion Models KI for Robotics Machine Learning

Research Highlight

SphereSR: 360º Image Super-Resolution with Arbitrary Projection via Continuous Spherical Image Representation

A new algorithm has been developed to generate omnidirectional images to super-resolve a low-resolution 360° image into a high-resolution image with arbitrary projection type via continuous spherical image representation....read more

Computer Vision Deep Learning KAIST Institute for Robotics Machine Learning Super-resolution Super-resolution with 360° Camera

Research Highlight

Reinforcement Learning Algorithm for Batch Chemical Process Control

Haeun Yoo, a member of Prof. Jay H. Lee’s group, has proposed a phase segmentation approach and modified deep deterministic policy gradient algorithm for batch process optimal control under uncertainty....read more

Actor-Critic AI Applications Batch process KAIST Institute for Artificial Intelligence Machine Learning Optimal control Reinforcement Learning

Research Highlight

Accelerated Material Design Framework using Uncertainty-Quantified Hybrid Machine Learning/Density Functional Theory Approach

Prof. Yousung Jung’s group has developed a new accelerated high throughput screening (HTS) method using uncertainty-quantified machine learning (ML) and density functional theory (DFT) that was applied to explore the Mg-Mn-O chemical space for photoanode application. Notably, the proposed HTS scheme required only 1.5% of the target chemical space for further DFT calculations, accelerating the entire process by > 50 times for the same discovery compared to the brute-force DFT-HTS done previously. This means an improvement of the screening performance (discoverability) by more than a factor of 2 compared to the conventional ML-based HTS approach....read more

CO₂ Conversion Density functional theory High throughput Screening Machine Learning Saudi-Aramco-KAIST CO₂ Management Center Uncertainty quantification

Research Highlight

Imaging with Events: High-resolution High-quality HDR Imaging using Event Cameras

New algorithms have been developed to generate non-blurry, high-resolution, high-quality, and high-dynamic range intensity images using sparse event streams from an event camera, which is a new imaging sensor able to capture visual information with low latency even under extremely low illumination....read more

AI for Cooperative Robots Computer Vision Deep Learning Event Camera KI for Robotics (KIR) Machine Learning Super-resolution Super-resolution with Event Camera

Research Highlight

Mole-bot: Development of bio-inspired embedded directional drilling robot

A research team led by Prof. Hyun Myung developed a bio-inspired digging robot called Mole-bot that simulates the mole's biological structure and behavior for resource exploration on Earth and other planets. In addition, a localization algorithm that can estimate the robot position and orientation in the underground environment was developed. The developed platform was verified through experiments on a testbed filled with soil....read more

AI KI for Robotics Machine Learning Mobile Intelligence for Vehicular Robots Robot Navigation SHM (Structural Health Monitoring) SLAM (Simultaneous Localization and Mapping) Swarm Robots Urban Robotics Laboratory

Copyright © 2015 KAIST MATRIX. All rights reserved.
291 Daehak-ro Yuseong-gu Daejeon, 34141, Republic of Korea
Partnered with KAIST Breakthroughs and KAIST Compass