• broken image
  • What's New

    Our preprint has been posted on arXiv.  Compositionality-Aware Graph2Seq Learning Takeshi D. Itoh, Takatomi Kubo, Kazushi Ikeda https://arxiv.org/abs/2201.12178
      Takatomi Kubo was appointed as associate professor of mathmetical informatics lab, NAIST, from Jan. 1st, 2022. https://isw3.naist.jp/Contents/Research/ai-04-en.html
    Our preprint has been posted on bioRxiv. https://biorxiv.org/cgi/content/short/2020.01.28.923953v1
  • Overview of the Project


    Our projects aim to unveil the cognitive processes in program comprehension as the unique higher cognitive function.

    It is mainly consisted of three approaches: brain activity analysis, gaze behavior analysis, and modeling.

    As the new projects, we are introducing other modalities, including ECG (not EEG).
    Please stay tuned!!

    fMRI scanner

    Brain Activity Analysis

    We are aiming to unveil the mechanism of neural processing during program comprehension with an fMRI scanner.

    broken image

    Modeling Approach

    We are modeling the process of program comprehension.
    Currently, we are using existing attention models for source code like code2vec, etc.
    We will evaluate the validity of models through comparison with the gaze behavior of human subjects.

    Tobii TX 300

    Gaze Behavior Analysis

    We are conducting experiments with an eye tracker to elucidate attended components in source code.

  • Members

    Takatomi Kubo

    Unit leader

    Yoshiharu Ikutani

    brain activity analysis (brain decoding)

    machine learning for codes (graph representation learning),

    contrasting gaze behavior and

    attention models

    Kiyoka Ikeda

    gaze behavior analysis, contrasting gaze behavior and

    attention models


    Toyomi Ishida


    Hayato Tsubaki

    machine learning (compositionality aware model)

    Shahoor Kausmally

    brain activity analysis

    Yurina Wada

    gaze behavior and/or ECG

  • Collaborators


  • Publication List


  • Outline of NAIST Summer School 2018

    Staff member

    • Takatomi Kubo
    • (TA) Yoshiharu Ikutani 
    • (TA) Eri Nakahara

    • Contents

    Aug 6

    1.   9:20 - 10:00: Registration, etc.
    2. 10:00 - 10:50: Introduction (Kubo, Ikutani, Nakahara)
      • Overview of general brain decoding 
      • Overview of brain decoding for code comprehension 
      • Preparation of materials (including installation of Matlab, SPM, and The Decoding Toolbox)
    3. 11:00 - 12:20: MRI and SPM (Kubo)
      • (Functional) Neuroanatomy
      • Principle of MRI measurement
      • fMRI and SPM
    4. 13:30 - 14:50: Machine Learning (mainly SVM) (Kubo)
    5. 15:30 - 16:00: Brain decoding-1 Introduction (Ikutani) 
    6. 16:00 - 17:00: Brain decoding-2 SPM Demo (Ikutani, Nakahara)

    Aug 7

    1.   9:20 - 15:00: Brain decoding-3 Practice (Ikutani) 
      (10:30 - 10:40: short brak; 12:30 - 13:30: Lunch break)
    2. 15:00 - 16:45: Brain decoding-4 Discussion (Ikutani) 


    0. Laptops will be prepared (except for the case you have Matlab in your system).

    1.  fMRI Data (will be distributed)
    2.  Matlab
    3.  SPM
    4.  The decoding toolbox
    5.  MRIconvert
    6.  Codes (will be distributed)


    • Textbook
    1. MATLAB for Neuroscientists (Second Edition) https://www.sciencedirect.com/science/book/9780123838360#ancp4
    2. Statistical Parametric Mapping
      (PDF) http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/
    • Web
    1. Matlab introduction https://jp.mathworks.com/help/pdf_doc/matlab/getstart_ja_JP.pdf
    2. SPM http://www.fil.ion.ucl.ac.uk/spm/
    3. Neuroanatomy of the Brain http://www.neuroanatomy.ca/