I am a Master's student in Computer Science at the University of Washington. I am working on multimodal semantic mapping with autonomous mobile robots under the supervision of Dr. Andrzej Pronobis. My academic interests span the area of robotics perception, machine learning, and probabilistic modeling.

Sep. 2017 - Jun. 2018

University of Washington

Sep. 2017 - Jun. 2018

M.S. in Computer Science

Jun. 2016 - Present

Research Assistant

Jun. 2016 - Present

Robotics and State Estimation Lab

Investigating machine learning methods that enable mobile robots to obtain the semantic understanding of their environment. Our approach is probabilistic and hierarchical, and works in real-time. Supervised by Dr. Andrzej Pronobis and Prof. Rajesh Rao.

Sep. 2015 - May. 2016

Research Assistant

Sep. 2015 - May. 2016

Integrated Brain Imaging Center

Developed a statistical analysis pipeline for fMRI data to discover a novel biomarker for Parkinson's disease, and introduced GNU Make to the brain research to enhance the reproduciility of experiments. Supervised by Dr. Tara Madhyastha.

Jan. 2015 - Sep. 2015

Research Engineer Intern

Jan. 2015 - Sep. 2015

Hymena

Invented an algorithm that detects personal information in hand-written documents, and developed a native scanner application that parses information in documents, then encrypts and uploads the structured data to cloud in just a few clicks. (Patented)

Fall 2017

CSE446 Machine Learning

Fall 2017

Teaching Assistant

Semi-Supervised Spatial Knowledge Transfer with Deep Generative Models

Kousuke Ariga

Semi-supervised spatial knowledge transfer improves a place categrization ratio (office, corridor, etc.) in a building by using unlabeled observations obtained in other buildings. Life-long place semantics learning, in short.

Senior Thesis

  • Training Deep Probabilistic Models for Semantic Mapping with Mobile Robots

    Kousuke Ariga, Kaiyu Zheng, Andrzej Pronobis

    Generated a large range-finder based dataset from both real world and randomly furnished simulated world. The dataset includes odometry, sensor readings, estimated poses, local representation of robocentric environment, etc.

    Undergrad Research Symposium

  • Generative Models for Missing Value Completion: Probabilistic PCA vs. VAEs

    Kousuke Ariga

    Compared two generative models: conditional variational auto-encoder (CVAE) and probabilistic principal component analysis both theoretically and empirically on missing value completion task. CVAE outperformed completion task by far.

    CSE446 ML Final Project

  • BOLD Variability and its Correlation to CSF Markers of Neurodegeneration in Parkinson Disease

    Ji-Min Park, Kousuke Ariga, Thomas Montine, Thomas Grabowski, Tara Madhyastha

    Discovered a correlation between BOLD variability and CSF markers of neurodegeneration of Parkinson Disease (PD). It potentially becomes a novel non-invasive method to observe the progression of PD.

    Undergrad Research Symposium
  • Shu Guru

    Kousuke Ariga, Rio Hoshi, Artem Zhabko, Jackson Cannon

    We developed a robotic system that supports busy shoe stores. A robot goes to back storage and retrieve shoe boxes that were ordered via mobile app. We used ROS framework and programmed the retrieval motion by demonstration.

    CSE481C Capstone Project
  • Research Onepager

    Kousuke Ariga

    I often forget what I read in which paper. This project is an attempt to make them trackable by summarizing a paper in a single slide and making them browsable from anywhere on the Web.